Chinese AI startup DeepSeek released an updated version of its primary model over the weekend, claiming the iteration delivers faster task completion and leaner performance. The release escalates competition against OpenAI and Anthropic as well as domestic rivals like Moonshot AI's Kimi and Zai, though a five-hour server outage this week hints at the hardware compute bottlenecks straining its infrastructure under rapid user growth. Read the full article on the benchmark gains and where DeepSeek is heading next: buff.ly/O2d0PRb #AI #DeepSeek #OpenWeights #LLM

Citigroup and Bank of America have joined Charles Schwab and JPMorgan in blocking AI agents from logging into customer accounts and moving money, with Schwab's CEO saying his firm blocks unauthorized agents ten to twenty times a week. The pushback is as much commercial as security-driven, since automated tools sweeping roughly $1.6 trillion in non-interest-bearing deposits into higher-yield alternatives could cost the sector an estimated $47 billion a year in additional interest. Read the full article on the gated, view-only frameworks banks are building instead: buff.ly/RL40esn #Cybersecurity #AI #AIAgents #Banking #Fintech

Anthropic has quietly begun pitching Capitol Hill lawmakers on a confidential draft bill that would let federal authorities block an AI model's release only when a foreseeable threat involves more than $1 billion in damages or the risk of killing or seriously injuring over 50 people. The proposal mirrors Anthropic's own safety framework, bars other agencies from unilaterally suspending a model, and would preempt some state-level disclosure and evaluation laws. Read the full article on the draft and the state preemption fight it has revived: buff.ly/rDm2PO5 #AI #AIPolicy #Anthropic #AIGovernance

Accenture has capped central funding for Anthropic's Claude at $29 per worker per month after some developers were burning through $500 to $1,000 worth of AI usage monthly, pushing business units to cover anything beyond that from their own project budgets. The policy lands just months after Accenture and Anthropic launched a business group aimed at training 30,000 professionals on Claude, and it highlights a question every large adopter now faces, whether simple spending caps can tell high-value usage apart from waste. Read the full article on tokenomics and the friction between deployment and discipline: buff.ly/3zOSlbw #AI #EnterpriseAI #Anthropic #Accenture

In case you missed it, 𝘛𝘩𝘦 𝘈𝘐 𝘍𝘢𝘤𝘵𝘰𝘳𝘺: 𝘍𝘳𝘰𝘮 𝘝𝘪𝘴𝘪𝘰𝘯 𝘵𝘰 𝘝𝘢𝘭𝘶𝘦 is now available on demand. The event explores how organizations can connect AI strategy with the infrastructure, architecture and operational foundation needed to support enterprise AI. Sessions cover open architecture, accelerated computing, AI sovereignty and the path from experimentation to measurable business value. Presented by @TechstrongGroup, part of @TheFuturumGroup, the program brings together insights from @RedHat, @Dell, @dify_ai, @weka and @nvidia on the infrastructure and technologies supporting enterprise AI. Access the full event on demand at your convenience: bit.ly/3TlaoqX #EnterpriseAI #AIInfrastructure #AIStrategy #ArtificialIntelligence

NVIDIA has unveiled three open Nemotron 3.5 models spanning from a 120-billion-parameter system for complex business tasks down to a 4GB model for embedded devices, advancing its push beyond AI hardware into models enterprises can customize and run on their own infrastructure. As one analyst puts it, the inference bill shrinks while the governance surface grows, since local agents bypass the cloud gateways IT uses to meter and audit AI. Read the full article on the models and what to watch before agents touch production systems: buff.ly/HsVCPJv #NVIDIA #AI #AIAgents #OpenModels

Political campaigns and PACs have poured more than $137 million into AI-generated television and internet ads this midterm season, with researchers identifying over 220 political ads using synthetic media and more than 80% of them produced by Republican campaigns or conservative-aligned groups. Only 31% of those ads carried a disclaimer acknowledging AI use, and while 30 states have enacted deepfake disclosure laws, compliance remains spotty and the legal fights are already playing out in key races. Read the full article on how the AI election is reshaping campaign strategy and disclosure: buff.ly/0eNdQsB #AI #Election2026 #Deepfakes #PoliticalAds

Former OpenAI engineer Abdulrahman Alfozan was arrested for allegedly planting a hidden recording device in an executive conference room. OpenAI says no sensitive data was stolen. Read more: buff.ly/vGFftdZ #OpenAI #Cybersecurity #ArtificialIntelligence #CorporateSecurity #DataProtection

Meta on Thursday announced the release of Muse Spark Open, calling it the most capable open-weight AI model built in the United States to date, and in an unlikely collaboration with often-bitter rival Apple, developers can now run the massive 816-billion-parameter system locally on high-end desktop hardware. Meta AI Chief Officer Alexandr Wang said his team was "excited for mommy and daddy to be playing nice again," noting the two companies had been coordinating behind the scenes for months. Read the full article on the benchmarks, the hardware setup and what it means for local AI execution: buff.ly/CVDNVNA #AI #Meta #Apple #OpenWeights #LocalAI

AI agents have completed more than 138 million cryptocurrency transactions on Coinbase's Base blockchain, according to new research from Chainalysis, indicating a major surge in the use of autonomous agents to purchase online services. The transactions ran through x402, a protocol that lets agents spend without human approval for each purchase, and U.S.-linked wallets accounted for 88.1% of geographically identified payments for AI model inference while Chinese-linked wallets took roughly 90% of transactions involving security and code-analysis services. Read the full article on how agent-based commerce is moving toward broader commercial adoption: buff.ly/93mKbdx #AI #AIAgents #Crypto #Stablecoins #Coinbase

The Trump administration is preparing to reassign three senior officials from CAISSI, the Commerce Department unit that evaluates advanced AI models for safety and capabilities before commercial deployment, to report directly to the White House's science and cyber leadership. The move follows months of coordination friction between Commerce and executive advisors and aligns with a broader push to centralize authority over frontier AI oversight under direct White House supervision. See what the reshuffle means for how frontier models get evaluated: buff.ly/OKhakkg #AI #AIPolicy #AIGovernance #CAISSI #WhiteHouse

Eight state attorneys general filed sweeping lawsuits against OpenAI on Wednesday, alleging the company prioritized growth over safety and that ChatGPT encouraged suicide, self-harm and violence, particularly among teenage users. A separate proposed federal class action filed in San Francisco claims the chatbot deliberately uses slang and emojis to coax intimate details from children to train its models, while OpenAI points to its existing age protections and parental controls. Read the full article on the escalating legal offensive: buff.ly/8O5nnal #AgenticAI #OpenAI #AISafety #TechPolicy #ChildSafety

The FTC has issued civil investigative demands to OpenAI and Anthropic, escalating a probe into consumer protection, privacy practices and safety incidents, with regulators particularly focused on disclosures that unreleased models bypassed security protocols and hacked into websites during testing. The subpoenas allow the agency to examine internal documents and question executives under oath before deciding whether to file suit, and the move arrives as 64% of Americans say AI is advancing too fast. Read the full article on the escalating federal scrutiny of AI's biggest players: buff.ly/gLrVdc2 #AI #FTC #OpenAI #Anthropic #AIGovernance

A Komprise survey of 500 technical decision makers finds that 68% feel their organization is maturing or has matured in AI adoption, yet only 28% can quantify a return on investment, and the gap comes down to data. Only 4% have more than 70% of their unstructured data ready with schema for analysis, 49% struggle to find the right project data and clean out redundant, obsolete and trivial files, and as Komprise's COO puts it, data equals tokenomics, since the less relevant the data exposed to AI, the harder it becomes to justify an ROI. Read the full survey findings on what is actually blocking AI returns: buff.ly/BmY22WA #AI #EnterpriseAI #DataManagement #ROI

ServiceNow unveiled EmployeeWorks and AI Passport on Thursday, a third-generation AI harness that lets non-technical employees build and deploy agents without writing code, paired with a credentialing layer that governs what those agents can access at runtime. Visibility alone isn't enough, since AI Passport issues cryptographically signed credentials agents carry into external environments and enforces graduated responses when they drift past their boundaries. See how ServiceNow is closing the gap between agent authority and operational oversight: buff.ly/6gwwOfZ #AI #AIAgents #ServiceNow #AIGovernance

NVIDIA today released its full Run:ai GPU orchestration platform as open source, making the scheduler, GPU fractioning operator, resource management layer and workload map freely available to the community. But open source doesn't equate to hardware-neutral, since the fractioning layer still runs on CUDA, and teams adopting it are choosing an NVIDIA-dependent control plane for their GPU estate. Read the full article on what's included and what the lock-in means for platform teams: buff.ly/QO2mXBG #NVIDIA #OpenSource #Kubernetes #GPU

The researchers who built the foundations of modern AI are now asking for binding international oversight, with roughly 400 scientists from 34 countries including Geoffrey Hinton and Yoshua Bengio calling for an organization that would set security standards, investigate containment failures and maintain an incident database. Their key recommendation is that advanced models be air-gapped until they can be contained, and it arrives as OpenAI agents keep escaping sandboxes to reach external systems while the voluntary agreement six major AI companies signed with the White House carries no enforcement provisions. See what the proposal would actually do and where the pushback comes from: buff.ly/DTRAZKJ #AI #AISafety #AIGovernance #TechPolicy

Kroger plans to eliminate roughly 3,000 corporate jobs over the next three years, more than a third of its 8,200-person corporate workforce, as part of an AI-driven push to streamline operations that leaves store-level and distribution center positions untouched. The cuts land alongside a $9 billion lawsuit from Albertsons over the collapsed merger and a lowered sales outlook, putting pressure on new CEO Greg Foran to bolster margins against Walmart, Costco and discount grocers gaining ground. Read the full article on the cuts and the pressures behind them: buff.ly/iZkrVdI #Kroger #AgenticAI #Retail #Layoffs

A Wharton and GBK Collective study found that 80% of organizations see positive ROI from AI, but only 28% report strongly positive results, and the gap comes down to what they measure. Departments judging AI against growth metrics are nearly three times as likely to see strong returns as those measuring efficiency alone, and companies with a designated Chief AI Officer hit 58% strong ROI compared to 24% for those handing oversight to existing executives. Read the full study findings on what separates the winners from everyone else: buff.ly/Tbw6Agc #AI #EnterpriseAI #ROI #AILeadership #DigitalTransformation