CTO Emerging Technologies, Innovation, Strategy, Startups, follow me for all of the above. Father, husband enjoying each moment as it happens.
McLean, VA
Joined April 2009
- Tweets12.5K
- Following969
- Followers10.6K
- Likes3.3K
AI is changing software development from two-week sprints to daily sprints. Developers are becoming orchestrators of AI agents rather than just writers of code. But faster development creates a new challenge: building what customers actually need, not just what AI makes easy. #AI #Startups buff.ly/gFwRh3N
What if AI could decode the language our cells use to communicate? MIT researchers developed an AI model that traces the signals guiding cells into tissues and organs. A breakthrough that could transform stem cell engineering, disease research, and regenerative medicine. #AI #Biotech buff.ly/Udk4Mqn
AI agents are raising a troubling new cybersecurity concern. OpenAI reports that one agent disabled its safety monitoring and then lied about it, while another suppressed security alerts. When AI can bypass its own safeguards, who's watching the watchdog? #AI #Cybersecurity buff.ly/0S1yTCK
AI transformation isn’t just about automating the work we already do.
Every major technology shift starts with doing old things faster. The real disruption comes later, when it enables things that weren’t possible before — and that no one had imagined.
A smart read from @benedictevans:
buff.ly/Nb39zwf buff.ly/Nb39zwf
18 months ago, AI could barely start a simulated cyberattack.
UK government testers just found it can now complete nearly half of one - a sequence that takes a skilled human 14 hours - for about $87.
No model has finished the whole thing yet.
The rate of change is the number that matters.
buff.ly/UVONaIV
#CyberSecurity #AI
Avo Reid retweeted
‼️ BREAKING: OpenAI was hacked by an Anthropic model. A HEIF photo uploaded to OpenAI's public support forum triggered a bug in the site's image decoder, led to code execution on the forum, and, through a second flaw in OpenAI's own login, ended with a pull request in OpenAI's internal GitHub.
The forum runs Discourse, the off-the-shelf software behind countless community sites. Discourse was still shipping an old copy of libheif, the library that decodes iPhone-style photos. The bug in it had already been fixed upstream.
But the fix was never labelled a security fix, so nobody treated it as urgent.
Hacktron's researchers uploaded a HEIF image and got their own code running on community[.]openai[.]com.
Then came the second bug, in OpenAI's own single sign-on, the "log in with OpenAI" button the forum uses. It turned that forum foothold into the actual ChatGPT and Codex accounts of people who had signed in there. OpenAI employees among them.
And a ChatGPT account is no longer just a chatbot. Through Codex, users wire in Gmail, Outlook, Drive, Slack, GitHub.
To prove the access was real, they used one employee account to have Codex open a harmless pull request in OpenAI's internal repo. They say they read no sensitive code.
OpenAI patched the SSO flaw roughly 14 hours after the report and paid a $6,500 bug bounty.
The team says Anthropic's Opus 4.8 found the libheif bug, and Opus 5 turned it into a working exploit.
Slack, Meta, GitHub Ent, Rails, Next.js, ImageMagick, and many more were also vulnerable and compromised by the same team of researchers.
Everyone says AI should slow down.
But what does “slow” actually mean — and who gets to decide?
A sharp look at the ambiguity behind calls to pause AI progress.
buff.ly/fmPo5y2
#AI #TechPolicy #ArtificialIntelligence
The software engineer of the AI era won’t be the person writing the most code.
They’ll define the architecture, set the guardrails, protect the data, and direct AI to do the rest.
The backlog is dead.
buff.ly/ahRqy2N
#AI #SoftwareEngineering #Startups #FutureOfWork
The Argument Against AI Writing Is At Least 2,400 Years Old feld.com/archives/2026/08/ai…
OpenAI agents were told to pass a test.
They escaped a sandbox, found a weakness, stole passwords, shared notes with each other, and broke into Hugging Face.
The scary question isn’t whether the AI “intended” to do it.
It’s that nothing stopped it in time.
As agents become more autonomous, prompts won’t be enough. Control, containment, monitoring, and layered guardrails become the real security model.
tomtunguz.com/openai-hack-ai…
#AI #AIAgents #AISecurity #AgenticAI #Cybersecurity #OpenAI
The next operating system may not be built for humans.
It may be built for AI agents.
As agents become autonomous, distributed, and always-on, they’ll need an infrastructure layer to coordinate resources, identity, communication, and human handoffs across devices and networks.
That’s not just an app stack. It starts to look like a new operating system for the agentic era.
buff.ly/qeHiK2A
#AI #AgenticAI #AIAgents #OperatingSystems #FutureOfWork #TechStrategy
China’s spy satellite story is really a rare-earth story.
A powerful radar eye in space depends on gallium nitride—and China dominates the gallium supply chain behind it.The next space race may be won before launch.
buff.ly/FXLYkC8
#SpaceTech #China #NationalSecurity #SupplyChain
How @supabase became essential infrastructure for the AI era: great developer experience, open-source roots, disciplined scaling, and relentless shipping.
Strong lessons from @kiwicopple via @firstround:
ift.tt/D5kFP8Y
#AI #DevTools #Startups
The AI frontier is becoming a repeating cycle:
Closed models create the breakthrough.
Open models catch up, cut the cost, and commoditize it.
Then the frontier moves again.
Open source may not need to lead the race to reshape the economics of AI. Competition is already forcing faster innovation and lower prices.
Via Tom Tunguz: [buff.ly/RzTYAmf](buff.ly/RzTYAmf)
#AI #OpenSourceAI #LLM #ArtificialIntelligence #TechStrategy
The AI inference market is splitting apart—just like databases did.
One category became relational, document, graph, vector, time series...
Now inference is fragmenting into latency tiers, multimodal, and edge. Different workloads, different stacks.
A $100B market looking for its Oracle, Mongo & Snowflake. buff.ly/61fQMhC
#AI #inference #LLM
Five Eyes agencies warn AI could bypass today’s cyber defenses in months—not years. Attackers are moving faster, and organizations can’t afford to wait. Modernize security, test defenses, and prepare now. #AI #Cybersecurity #InfoSec #FiveEyes #CISO #CyberRisk buff.ly/Le79GcB
The Census Bureau’s 2026 AI adoption data cuts through the hype.
Across all U.S. businesses, AI use is still only around 17–20%. But employment-weighted adoption rises to 32%, meaning the average employee is increasingly working somewhere that uses AI.
The real divide is by company size and sector:
Large firms are adopting faster.
Information and finance are ahead.
Retail and smaller businesses are still catching up.
The takeaway: AI adoption is not universal yet. It is concentrated, uneven, and highly dependent on industry, company size, and business function.
That makes Census data a powerful map for where AI demand is real versus where the market is still mostly noise.
#AI #EnterpriseAI #AIEconomy #Census #GenAI
buff.ly/CdTJRCU
By 2030, AI data centers will guzzle 9.3 trillion liters of water — equal to the annual needs of 1.3 billion people. And making AI more efficient might actually make it worse, since cheaper AI means more usage. The UN says we're solving one problem while creating others. buff.ly/ABDQiRZ
The next phase of enterprise AI won’t just be model adoption. It will be model substitution.
As frontier models get more expensive and open-source models become “good enough” for more use cases, AI buyers are starting to route work to the cheapest capable model.
That changes the economics:
Costs don’t simply fall.
Usage expands.
The winners will be the companies that build intelligent routing, benchmark performance by task, and optimize AI spend like cloud infrastructure.
AI strategy is becoming unit economics strategy.
#AI #EnterpriseAI #GenAI #OpenSourceAI #ArtificialIntelligence
buff.ly/ben3PZk
The real AI disruption may not be the model.
It may be the stack.
AI-native operating systems and tools are forcing a deeper rethink of software design itself, not just adding smarter features to old workflows.
#AI #AINative #DevTools #FutureOfComputing buff.ly/1xy52QK
