Bhaskar Pandey (@bhaskar_author) header
Bhaskar Pandey (@bhaskar_author) profile picture

Data, governance & political I help teams turn complex ground realities into decisions, operating systems and execution. Consulting, advisory & Execution

Chennai, India
Joined May 2015
Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)Photo by Bhaskar Pandey (@bhaskar_author)
From Indian mobile-number research to a 22-circle telecom data + GIS prototype: numbering catalogs, Python pipelines, spatial models and network simulations. Next milestone: independent validation, not more features. What datasets or benchmarks should we test against?
1
19
What if academic research could become real-world technology products? We're working on exactly that through our Independent Research-to-Product initiative. Our first AI product is taking shape: Beyond Likes. Here's what we're building. 🧵 #BuildInPublic #AIResearch
1
1
159
Our roadmap is straightforward: Research → Engineering → Independent Validation → Real-World Pilots → Commercialization. We're also open to conversations with early-stage investors interested in research-driven technology. But first, we want evidence #Startups #VentureCapital
1
3
Want to be part of building Beyond Likes? We're inviting early testers, technical collaborators, researchers, and potential pilot partners. Reply "BETA" or DM me with your background. Let's test the technology before asking the market to trust it. #BuildInPublic #AIStartups
14
Introducing Beyond Likes. An AI-powered discourse intelligence system designed to: • Evaluate conversation quality • Identify unconstructive interactions • Suggest better ways to communicate • Help communities improve discussions #MachineLearning #AIProducts
6
India’s next power-sector reform may have less to do with building another power plant and much more to do with building a better operating system for the ones we already have. As of August 25, 45 coal-fired power plants were reportedly carrying critically low fuel inventories. Coal stocks across plants had fallen to roughly 30.95 million tonnes around 10 days of operational supply after monsoon disruption affected mining and transportation. Now compare that with December 2025. India had approximately 53.5 million tonnes of coal available at power plants, enough for around 17 days at the prescribed requirement level. This is where the governance discussion becomes interesting. Because India does not fundamentally lack: Coal. Railways. Power plants. Engineers. Data. Or institutions. What we often lack is a sufficiently integrated system connecting all of them in real time. A coal shortage at a thermal plant rarely begins at the plant. It may begin hundreds of kilometres away: Mine production falls. Rain disrupts loading. A rake isn't available. A route gets congested. Demand unexpectedly rises. Inventory falls faster than forecast. Maintenance gets postponed. And suddenly something that looked like a logistics problem becomes an electricity-security problem. This is why India's next infrastructure leap should be an Energy Operations Grid. Imagine one national operational layer continuously combining: → mine-level production → railway rake availability → plant-wise coal inventories → weather forecasts → reservoir positions → renewable generation forecasts → state-level electricity demand → transmission constraints → maintenance schedules → spot-market electricity prices Then predictive models ask a very simple question: Which plant is likely to face stress 7, 14 or 30 days from now? Not after the stock reaches critical levels. Before. That changes governance from crisis response to anticipatory administration. A railway rake could be redirected earlier. Maintenance could be rescheduled intelligently. Generation could shift geographically. Gas/hydro/storage capacity could be prepared. States could purchase power before spot prices spike. Coal dispatch could be prioritised based on projected grid importance rather than merely today's inventory. And renewable intermittency becomes another input into the same operating model rather than a separate policy conversation. India has spent decades constructing physical infrastructure. The next competitive advantage will come from creating the digital coordination infrastructure sitting above it. This applies far beyond electricity. Food grain movement. Fertiliser. Water. Ports. Urban transport. Healthcare supplies. Disaster management. The Indian state already generates extraordinary amounts of operational data. The question for the next generation of governance is whether that data remains inside departmental dashboards or becomes a real-time decision system. We often debate government capacity by asking: “How much infrastructure did we build?” Increasingly, we should also ask: “How intelligently are we operating the infrastructure we already built?” That may become one of the defining governance questions of India's next decade. @MinOfPower @CEA_India @RailMinIndia #India #Governance #EnergySecurity #Infrastructure #PowerSector #DigitalIndia #PublicPolicy
1
146
India cannot settle delimitation with arithmetic alone. The debate returning from the Southern Zonal Council is being framed as North vs South. That framing is politically convenient and constitutionally inadequate. There are actually two democratic claims colliding. A voter in Bihar or Uttar Pradesh should reasonably ask why an MP there represents far more citizens than an MP in some southern states. “One person, one vote” cannot remain permanently detached from population. But Tamil Nadu, Kerala and other lower-fertility states have an equally serious question: if a state implemented population-control policies successfully, improved human development and reduced fertility, should that success eventually reduce its political influence in Parliament? That would create a dangerous incentive structure. India froze interstate parliamentary seat allocation decades ago precisely because demographic transition was occurring at very different speeds. The 84th Constitutional Amendment extended that protection until the first census after 2026. Now the bill has come due. And simply increasing Lok Sabha seats and distributing them strictly according to population does not solve the political problem. It merely converts a demographic imbalance into a federal one. The better question is: How do we restore greater equality between individual voters without destroying equality between constituent states? India may now need a broader federal bargain: 1. Expand the Lok Sabha so faster-growing states gain representation without requiring large absolute losses elsewhere. 2. Consider degressive proportionality rather than mechanically applying pure population ratios. Larger states would receive more seats, but smaller/slower-growing states would retain a reasonable institutional floor. 3. Reconsider the Rajya Sabha’s federal role. In many federations, the upper chamber protects constituent units against domination purely by population. India’s Rajya Sabha itself remains heavily population-weighted. 4. Link delimitation with the fiscal conversation. Representation, tax contribution, Finance Commission transfers, population performance and development incentives cannot forever be debated in separate rooms. Most importantly, Parliament cannot treat delimitation as another majority-versus-opposition vote. Changing the relative political weight of Uttar Pradesh, Bihar, Tamil Nadu, Karnataka, Kerala, Maharashtra and other states potentially changes the architecture through which India itself is governed. The North should not remain permanently under-represented. The South should not be punished for demographic success. And neither side should be told that constitutional arithmetic alone settles the question. Delimitation is ultimately not about drawing constituencies. It is about renegotiating India’s federal compact for the next several decades. That deserves consensus substantially wider than whoever happens to command the numbers in Parliament today. #Delimitation #Federalism #IndianPolitics #Parliament #Democracy #Governance
1
41
Calling the “China shock” a myth risks making almost the same analytical mistake as blaming China for the entire decline of American manufacturing. Both stories are too simple. Yes, US manufacturing employment had been declining for decades before China joined the WTO in 2001. Automation, productivity gains, offshoring, the rise of services and structural economic change were already reshaping the labour market. But that does not mean the China shock was imaginary. The work of Autor, Dorn and Hanson showed something much more specific: American regions and industries with greater exposure to Chinese import competition suffered larger manufacturing job losses, weaker wages, lower labour-force participation and surprisingly persistent adjustment problems. So these two statements can both be true: America was already deindustrialising in employment terms. And: Chinese import competition accelerated and concentrated that pain in particular communities. The distinction matters because national averages hide local economic trauma. A 1% employment shock spread evenly across an economy is not socially equivalent to a 1% shock concentrated in a handful of industrial towns. There is another distinction we often miss: manufacturing employment ≠ manufacturing capacity. US manufacturing output and value-added could continue rising even while factories employed fewer workers. Productivity, automation and movement toward higher-value activities can produce exactly that outcome. And this is where @MichaelRStrain’s argument becomes far more relevant than the headline suggests. The real lesson of the China shock is not “free trade failed” or “protectionism works.” It is that aggregate gains do not automatically compensate concentrated losers. That lesson matters enormously for AI. Artificial intelligence may increase productivity, corporate profits and national GDP while simultaneously destroying demand for specific professions, reducing wages in certain occupations and concentrating disruption in particular cities or worker groups. The economy can win while communities lose. So the policy question should not be: How do we stop economic change? It should be: How do we measure who bears the transition cost, how quickly disruption occurs, and whether enough of the gains are used to help those people adjust? That is where both protectionists and free-market absolutists often fall short. The China shock was not a myth. The myth was believing that national economic gains would automatically translate into painless local adjustment. Important debate from @bsindia and @MichaelRStrain. #ChinaShock #Trade #Manufacturing #AI #Economics #IndustrialPolicy
#Opinion | The myth of the China shock The decline in #US manufacturing jobs began decades before #China’s #WTO entry, writes @MichaelRStrain. He argues the bigger challenge is helping workers and regions adapt to #trade and technological disruption, rather than restricting economic change. mybs.in/2g8WUOw
1
37
This headline sounds like a privacy story. It is actually an architecture-of-power story. If a future ChatGPT can continuously see your screen, listen to meetings, read messages, follow documents and remember context, the real question is not simply: “Did the user consent?” Because most of what appears on your screen does not belong only to you. A client email contains somebody else’s data. A meeting contains other people’s voices. A corporate dashboard may contain confidential information. A hospital file contains sensitive health data. A board document may contain price-sensitive information. A lawyer’s screen may contain privileged communication. So one person clicking “Allow” cannot automatically legalise everything an AI agent may subsequently observe. And this is where Sam Altman’s broader vision runs directly into modern privacy law. AI systems become more useful with maximum context. But GDPR-style regulation is built around data minimisation, purpose limitation, lawful basis and storage limitation. That is a structural conflict. The US is more fragmented FTC enforcement, sector-specific laws, state privacy laws and different call-recording rules. Europe will be much tougher because GDPR, ePrivacy and the AI Act can operate simultaneously. India is entering its own DPDP regime. Japan, South Korea, Singapore, Australia, Brazil and China all have different consent, localisation, sensitive-data and cross-border-transfer requirements. So there may never be one universal version of an “always-aware” AI assistant. The product may eventually need: • granular permissions for screen, microphone, email and files • visible recording indicators • meeting-participant consent • local/on-device processing • short raw-data retention • separate controls for memory and model training • jurisdiction-specific data residency • enterprise audit logs • automatic exclusion of sensitive applications The decisive distinction will also be: capture ≠ processing ≠ storage ≠ memory ≠ training An AI listening locally, extracting action points and deleting raw audio is fundamentally different from storing entire meetings indefinitely. So the question raised by this @bsindia report is much bigger than “Will ChatGPT watch us?” It is this: Can humans legally delegate their eyes and ears to a permanent AI agent when those eyes and ears continuously encounter information belonging to other people? That may become one of the defining privacy battles of personal AI. @OpenAI @sama #ArtificialIntelligence #DataPrivacy #AIRegulation #OpenAI #GDPR #Technology
#OpenAI CEO #SamAltman has predicted a significant shift in how humans could interact with #AI assistants, saying a future version of #ChatGPT could watch a user's computer screen, meetings, and record calls all the time. Details here 👉 mybs.in/2g8WUAU #ArtificialIntelligence
2
2
70
The headline is that PM CARES received fewer donations and spent only ₹87.85 lakh in FY2024-25. But that is not the real governance question. The Fund closed the year with ₹8,452.06 crore. Around ₹7,846 crore was held in fixed deposits, generating roughly ₹469 crore in interest almost as much as the ₹479.96 crore received in fresh contributions. More importantly, about ₹324.65 crore came back from implementing agencies. None of this, by itself, proves wrongdoing. An emergency fund can legitimately maintain reserves. An implementing agency can legitimately return unspent money. But once a national emergency institution accumulates ₹8,000+ crore, the standard of scrutiny has to move beyond: “the accounts have been audited.” The questions should be administrative, not rhetorical: 1. What is PM CARES' reserve-adequacy policy? What calculation determines how much should remain invested and how much should be deployed? 2. What triggers deployment? What objective criteria determine when a flood, cyclone, epidemic, infrastructure emergency or other disaster qualifies for PM CARES intervention? 3. What exactly is the ₹324.65 crore returned by implementing agencies? Agency-wise and project-wise, what was sanctioned, released, utilised and returned and why? 4. Where are the utilisation certificates and outcome reports? Financial audit tells us where money sits. It does not tell us whether equipment worked, infrastructure became operational, beneficiaries were reached or expenditure delivered value for money. This distinction matters. The question is not “Where did ₹8,452 crore disappear?” The published accounts show that the money exists. The real question for @PMOIndia is: What publicly disclosed governance framework determines when this ₹8,452-crore emergency corpus should accumulate, when it should deploy, and how its outcomes should be measured? That is not an allegation. It is the accountability architecture any institution of this size should be able to answer. Important reporting by @IndianExpress and Harikishan Sharma. #PMCARES #Governance #PublicAccountability #Transparency
Prime Minister’s Citizen Assistance and Relief in Emergency Situations Fund (PM CARES Fund) contributions fell to Rs 479.96 crore in financial year 2024-25, while its closing balance rose to an all-time high of Rs 8,452.06 crore. Spending from the Fund also fell to a five-year low of Rs 87.85 lakh, according to its latest audited accounts. Click on the 🔗 below to read the full article. indianexpress.com/article/in… Written by: Harikishan Sharma
1
14
This paper by Hyunwoong Pyun, shared by @EEJ_easternecon, raises a more interesting question than simply asking why young people participate less in politics: What if uncertainty itself changes the kind of politics young people respond to? The paper's model suggests that younger citizens may struggle to identify which policies best serve their long-term interests because their own futures are far less predictable. A 55-year-old generally has a clearer idea of income, assets, profession, taxation and retirement needs. A 22-year-old may not yet know: whether they will have a stable job, what income bracket they will occupy, which city or country they will live in, whether they will own property, or even whether their present profession will survive technological change. That uncertainty can reduce the perceived benefit of choosing between competing long-term policies. But I think the political implication goes one step further. The danger may not simply be that uncertain young people withdraw from politics. It may be that uncertainty changes what kind of politics is easiest to sell to them. When the economic future is difficult to calculate, policies whose benefits arrive 10 or 20 years later become harder to evaluate. Meanwhile, things with an immediate emotional or material payoff become much easier to understand: identity, anger, nationalism, cash transfers, subsidies, symbolic recognition, cultural conflict, charismatic personalities. This matters enormously for India. We already have millions of young people navigating an uncertain path from: education → examination → employment → economic mobility. Degrees do not guarantee capability. Capability does not guarantee a good job. Competitive examinations may mean years of waiting. AI is beginning to disrupt precisely the entry-level white-collar occupations many graduates expected to enter. In such an environment, political volatility should not automatically be read as apathy. A young citizen can vote, protest, become intensely political around one issue, abandon a party the next election, and disengage again afterward. That may be less about ideological inconsistency and more about an unstable economic future producing unstable political preferences. One important caution: Pyun's study is a theoretical simulation, not empirical proof of how young voters actually behave. But the mechanism is worth taking seriously. Perhaps the deeper democratic question is no longer: “Why are young people disengaging?” It is: “What happens to democracy when an entire generation cannot confidently predict its own economic future?” #Youth #Politics #PoliticalEconomy #Democracy #Employment #India #PublicPolicy
1
176
Three curves explain India's youth-employment problem better than most Left-vs-Right arguments do. 1. Access to credentials has expanded rapidly. 2. Actual learning and productive capability have grown much more slowly. 3. The supply of desirable, productive jobs has grown slower still. That widening distance is where the crisis sits. @bsindia and Nitin Desai are right to question whether India's education system is producing the outcomes that decades of enrolment expansion promised. But reducing the problem to “spend more on education” would miss half the story. India has dramatically increased the number of graduates. Yet millions enter a labour market every year where the number of stable salaried opportunities is nowhere close to the number of people carrying degrees. At the same time, saying this is merely a “skills problem” is equally convenient. If five million graduates are entering the system while substantially fewer are being absorbed into productive employment, teaching everyone another certification course does not manufacture the missing jobs. There is another uncomfortable layer we rarely discuss. Unemployment and employment are not opposites in the way political debates pretend. A graduate supported by their family can remain “unemployed” for two years preparing for UPSC, SSC, banking or another preferred job. A poorer graduate may immediately accept delivery work, informal sales or low-paid employment. Statistically, the second person is employed. Economically, that does not necessarily mean the system worked better for them. Then add examination leaks, delayed recruitment, unfilled vacancies and the enormous coaching economy. We aren't merely facing a jobs crisis. We are facing a credibility crisis in the entire education → examination → employment → mobility pathway. Nor should the answer become “schools instead of roads” or “manufacturing instead of services”. Education, infrastructure, firms, manufacturing, services, apprenticeships and trustworthy institutions have to work together. India's real failure is more fundamental: we expanded access to credentials faster than access to capability, and capability faster than access to productive employment. Until those three curves begin converging, both sides will continue finding statistics that prove their politics while young Indians continue living the gap between them. #India #Education #Jobs #YouthEmployment #Economy #PublicPolicy
#Opinion | #India’s rising number of educated young people has not translated into enough stable jobs, with education quality remaining a key challenge. Nitin Desai argues that the Centre and states will need to rethink spending, standards and management of education. mybs.in/2g8W2cG
1
22
This looks like a technical compliance update. It is actually the beginning of a much larger argument about authorship, provenance and who gets to certify digital expression. @the_hindu reports via @SahanaV_Journo that @AnthropicAI will introduce machine-readable markings in future Claude-generated content as part of compliance with the EU AI Act. At first glance, that sounds sensible. Synthetic media, deepfakes, automated propaganda and industrial-scale misinformation have made provenance infrastructure increasingly necessary. But there is an important distinction we should not lose: AI provenance is useful. Binary AI authorship is dangerous. Consider a very ordinary workflow. A journalist writes an original article and asks Claude to improve grammar. A researcher writes a paper and uses AI only to restructure two paragraphs. A lawyer drafts an argument and asks AI to simplify the language. An Indian author writes an original essay in Hindi and uses Claude only to translate it into English. If machine-readable signals appear in these outputs, what exactly have we established? That AI was involved? Or that AI authored the intellectual work? Those are completely different claims. And this distinction will matter far beyond social media. Universities may use AI-detection signals in academic-integrity proceedings. Publishers may use them while evaluating manuscripts. Employers may use them while assessing applicants. Courts may eventually encounter them while determining evidentiary provenance. Newsrooms may use them while assessing submitted material. A technical marker designed to indicate machine participation could very easily evolve into a social judgment about human originality. That would be a serious mistake. There is another institutional question. As companies such as Anthropic, Google, OpenAI and others develop watermarking and verification infrastructure, we should avoid a future where individual AI companies effectively become private authentication authorities deciding whether digital content is “AI-generated.” Provenance standards must therefore be interoperable, independently auditable and capable of representing degrees and types of AI involvement, rather than merely producing a red or green verdict. We will soon reach a point where asking whether something is “AI-generated” becomes almost meaningless. The better questions will be: Who created the underlying idea? What part did the machine perform? Was the content merely edited, translated or formatted? Was a synthetic image or factual claim generated? Can the chain of modifications be independently verified? The EU AI Act is pushing the industry toward provenance. That direction is understandable and probably necessary. But regulation must avoid turning “AI touched this” into “a human did not create this.” The future of authorship is unlikely to be human versus machine. It will increasingly be human with machine. Our standards need to become sophisticated enough to understand the difference. #ArtificialIntelligence #AIAct #Anthropic #Claude #AIGovernance #DigitalProvenance #TechPolicy #ResponsibleAI
As part of its measures to comply with the European Union’s AI Act, the Google and Amazon-backed Anthropic announced that Claude content in the future will carry a machine-readable marking to better enable AI content labelling. ✍️@SahanaV_Journo thehindu.com/sci-tech/techno…
2
66
The most interesting sentence in this entire @DarioAmodei–@GavinSBaker–@_sholtodouglas exchange isn't about extinction, open weights or even cancer. It's $500 million. The conversation began with an extraordinary second-hand claim about Anthropic potentially becoming the only private company left alongside governments. Sholto Douglas flatly denied it. Gavin subsequently accepted the correction. So that allegation should remain exactly what it is: unverified, disputed and not a fact on which to build an argument. But Gavin's underlying concern survives the disputed anecdote. Dario argues that Anthropic deliberately supports regulation that hurts frontier incumbents while protecting challengers. His exhibit A is California SB53: roughly speaking, the heavier framework applies to “large frontier developers” above $500M annual revenue. Except this is precisely where the governance problem becomes interesting. SB53 does not simply mean “below $500M = completely outside regulation.” California distinguishes between a frontier developer and a large frontier developer; some obligations concern frontier developers generally, while additional framework/governance requirements fall on the $500M+ group. More importantly: suppose a challenger grows from $450M to $550M. Its model didn't suddenly become dangerous because its sales team had a good year. Yet its regulatory position changes. For Anthropic-scale firms, evaluation teams, lawyers, safety researchers, red-team infrastructure and government interfaces are already fixed costs. For the company trying to become the next Anthropic, those same requirements can become a competitive transition tax precisely when it begins challenging the incumbent. That doesn't prove regulatory capture. But neither does putting a $500M exemption into legislation prove decentralisation. The better trigger is increasingly obvious: regulate demonstrated capability and externality, not corporate size alone. Bio-risk capability? Test it. Autonomous cyber capability? Test it. Ability to recursively conduct frontier R&D? Test it. Then make the evaluator institutionally independent, publish the methodology, permit adversarial review, create an appeals mechanism and make compliance progressively scale instead of appearing as a cliff. Dario's other argument deserves the same evidentiary standard. He says AI may help cure most human disease within ~5–10 years and simultaneously says AI companies should stop making empty promises and actually cure cancer. Good. Then let's measure the distance between prediction and delivery. Anthropic's current rare-disease initiative is funding researchers with up to $50,000 in Claude credits, working on areas such as disease-mechanism discovery, drug repurposing, variant classification, regulatory documentation and clinical-development bottlenecks. Anthropic itself explicitly acknowledges limitations including poor data, manufacturing constraints, safety testing and healthcare access. That's meaningful progress. It is not yet “curing most disease.” And perhaps that is the real lesson of this debate. Don't ask society to choose between trusting Silicon Valley and trusting regulators. Trust neither automatically. Build systems where claims can be tested, decisions challenged, incumbents contested and promises measured against outcomes. The danger isn't only AI becoming concentrated. The danger is allowing the people building the frontier, the people regulating the frontier or the people predicting the frontier to become their own source of validation. That is where governance should begin. @Anthropic #AIGovernance #AIRegulation #Anthropic #OpenWeights #CompetitionPolicy #AISafety #TechPolicy
1/2 Thanks Gavin for an especially thoughtful exchange. I don't usually spend much time on social media but I wanted to engage here because it really brings out the heart of an important conversation. First, on regulation, I think that “either concentrate it in the hands of a chosen few companies and politicians via regulation or distribute it widely” is a false choice.  I know that there’s a sort of Silicon Valley shorthand where regulation = regulatory capture = concentration of power, but I’ve always found this to be an overly simplified picture of the world.  Many people outside this bubble think of regulation as something that constrains corporate power and benefits ordinary people.  I don’t necessarily agree with that perspective either, rather I think it’s complicated and really depends on what the “regulation” consists of.  But in particular I think that those in the “regulation = regulatory capture = concentration of power” frame often underrate the decentralizing power of objective and fair institutional processes.  A crude analogy is that the formal court system can sometimes feel stuffy and elitist, but it does a much better job of defending the rights of vulnerable individuals than the alternative, mob justice.  At their best, institutions can vest power in ideas rather than people, and thereby decentralize that power. This is why Anthropic has always made its policy proposals very carefully.  We try very hard to make proposals that disadvantage (slow down) frontier AI companies while *advantaging* smaller competitors.  California’s SB53 (which we supported), and even the much-maligned SB 1047 (which we were ambivalent on), completely exempt any company below a certain amount of revenue or model training costs from being covered at all (it was $500M for SB 53, lower for 1047 but we objected to that).  More recently the testing process we’ve advocated for at CAISI and the White House involves more rigorous tests for frontier models than off-frontier models — something that differentially advantages challengers.  Similarly, the “Pacing the Frontier” letter envisions (or at least Anthropic’s preferred implementation of it envisions) modulating the pace of the very best models while not constraining those who are catching up.  This hurts the business interests of the frontier labs and helps challengers, including open-weights! Overall my view is that AI is *structurally* a technology that tends to concentrate power, for reasons that have nothing to do with regulation (more to do with the extreme implications of the scaling laws).  Open-weights do help some with this but are nowhere near a sufficient solution because they simply shift the concentration somewhat to those with the most compute and chips (which are roughly the frontier labs plus maybe hardware providers).  By contrast I think the right “rules of the road” can simultaneously (a) address AI’s cyber/bio/alignment risks, (b) institutionally constrain the power of the frontier AI companies, and (c) leave room for open-weights models while also addressing the specific risks that they bring. BTW I do not think that the events of the last few months have “failed to result in [my] preferred regulatory path”.  The approach that the Trump administration is reported to be taking — pre-deployment testing for frontier models, and also testing of open-weights models when they get closer to the frontier — is one that I am very supportive of, though of course I have to see the details to be sure.  I am also supportive of Demis Hassabis’ ideas around a FINRA-like entity.  This contrasts with six months ago when most of the industry was still pushing for preemption of all state regulation and no apparent federal approach either.
1
138
If surrogate advertising is the offence, don't investigate only the surrogate. Investigate the advertising ecosystem. The show-cause notices reportedly issued over the Vimal Elaichi campaign raise a legitimate accountability question. Celebrities should absolutely answer for endorsements. They accepted the contract, lent their reputation to the product and helped create enormous brand recall. Asking what due diligence they conducted is fair. But if enforcement ends with three famous faces, we will have missed the much larger issue. Because an actor did not wake up one morning and independently design a ₹crore advertising campaign. Someone owns the brand. Someone manufactures and markets the product. Someone designed the campaign. Someone approved the creative. Someone purchased television, outdoor and digital inventory. Someone decided which celebrities to hire. Someone repeatedly distributed the advertisement across television, cricket, YouTube, social media, billboards and other platforms. And ultimately someone benefits financially from keeping that brand permanently visible. Why is that entire chain less interesting than the celebrity standing in front of the camera? Surrogate marketing is hardly a phenomenon India discovered in 2026. For years we have watched restricted-product brands appear through elaichi, mouth fresheners, soda, packaged water, music, events and other extensions. The legal details differ from campaign to campaign, and merely sharing a brand name does not automatically prove surrogate advertising. That is precisely why regulators should investigate the commercial architecture, not merely produce celebrity headlines. If @FDA_MAHARASHTRA believes this particular campaign prima facie crosses the line, then the next questions should be unavoidable: Who commissioned it? What was the advertising objective? What percentage of the advertised extension's actual business justifies the scale of promotional spending? Which agency created the campaign? Who approved its media strategy? How much was spent? And if a campaign is ultimately held unlawful, what happens to its continuing distribution across broadcasters and digital platforms? This should also trigger a wider question for @jagograhakjago: Why examine one campaign and not conduct a category-wide audit? Apply the same transparent standard to pan-masala extensions, liquor-brand extensions, tobacco-linked branding, sports sponsorships, outdoor campaigns and digital advertising. Otherwise we risk creating headline enforcement: Three actors receive notices. Television debates their morality. Social media trends for two days. And the commercial machinery capable of producing the next surrogate campaign remains completely intact. This is not an argument for letting celebrities escape responsibility. Quite the opposite. Ask the endorsers: What did you know, what did you verify and what exactly did you agree to promote? But ask the corporation: Why was the campaign designed this way? Ask the agency: What behaviour and brand association was it intended to create? And ask the distribution ecosystem: Why should prohibited advertising continue generating reach and revenue once regulators determine it violates the law? Accountability should follow the money as well as the face. Because three actors may have delivered the message. But somebody manufactured the message, somebody financed it, somebody distributed it, and somebody ultimately profited from the recall it created. Regulation that stops at the celebrity isn't regulation of surrogate advertising. It is regulation of its most visible face. @the_hindu @FDA_MAHARASHTRA @jagograhakjago #SurrogateAdvertising #ConsumerProtection #Advertising #PanMasala #TobaccoControl #PublicHealth #BrandAccountability #MarketingEthics #CorporateAccountability #India
The Maharashtra Food and Drug Administration (FDA) has issued a show-cause notice to actors Shah Rukh Khan, Ajay Devgn and Tiger Shroff over their advertisement for Vimal Elaichi, claiming it prima facie amounts to "indirect or surrogate" promotion of prohibited pan masala and tobacco-related products. thehindu.com/news/national/m…
1
66
When a government has to cancel its own recruitment examinations, investigate the examination vendor, and confront allegations serious enough to keep thousands of aspirants on the streets, “this is only politics” stops being an adequate answer. The protests now confronting @HemantSorenJMM and @JharkhandCMO are not about one examination or one political party. They are about something more damaging: whether a candidate in Jharkhand can still trust the next recruitment examination. The distinction matters because not every allegation has been judicially proved. Take JSSC-CGL. The government has refused to cancel it, arguing that the process proceeded under court directions and appointments have already been made. The Jharkhand High Court had earlier declined to order the CBI investigation sought by petitioners. So it would be irresponsible to simply declare the examination “proven rigged.” But that does not end the scrutiny. NDTV has since reported details of an earlier CID investigation that allegedly found that 120 of 150 questions had leaked, before a subsequent investigation reached a different conclusion. If that report is authentic, there is an obvious public-interest question: What changed between the two investigations? The government should publish as much of that evidentiary trail as legally possible. Because transparency is particularly important when different investigative processes appear to produce radically different conclusions. Then consider JPSC. The government has now agreed to cancel the 14th JPSC examination along with two other JPSC examinations and investigate examinations conducted through the concerned vendor. That is not evidence that every accusation raised by protesters is correct. But neither is it evidence of a system functioning normally. And this is where @JmmJharkhand, its coalition partner @INCJharkhand_, and opposition @BJP4Jharkhand should resist turning another recruitment crisis into nothing more than partisan ammunition. A student preparing for a government examination may spend two or three years on it. Coaching fees. Rent. Travel. Books. Repeated applications. Foregone earnings. Family savings. When an examination collapses, that loss cannot simply be measured as another cancelled test date. The state has consumed a portion of someone's life. Cancelling examinations is therefore emergency treatment. It is not recruitment reform. Jharkhand now needs an institutional reset: Publish an exam-by-exam status report. Disclose vendor selection and due-diligence processes. Create independent technical audits covering question-paper custody, encryption, printing, logistics and result processing. Establish clear accountability when an examination fails. And consider compensation or fee protection when candidates are forced to repeat examinations because of proven administrative failures. The students have announced plans to gherao the Chief Minister's residence on August 20 and are demanding his resignation. They have every democratic right to make that political demand. But whether @HemantSorenJMM resigns is ultimately a political question. Whether Jharkhand can conduct a recruitment examination that young people trust is a governance question. And whoever governs Jharkhand after the next election will inherit that question if it remains unanswered. A government job examination should test the candidate. It should not become a test of how much uncertainty, litigation and institutional failure the candidate can survive. @the_hindu @HemantSorenJMM @JharkhandCMO @JmmJharkhand @INCJharkhand_ @BJP4Jharkhand #Jharkhand #JPSC #JSSC #ExamReform #PaperLeak #GovernmentJobs #Youth #Recruitment #Governance #Accountability #IndianPolitics
Students, protesting against alleged irregularities in Jharkhand recruitment examinations, on Saturday (August 15, 2026), accused the ruling JMM and the Congress of playing politics over their issues and announced that they would gherao Chief Minister Hemant Soren’s residence in Ranchi on August 20, demanding his resignation. thehindu.com/news/national/j…
1
49