Editor + founder @ReadTransformer, covering AI policy. Prev: @TheEconomist, @Protocol, @finimize
London
Joined October 2009
- Tweets17.3K
- Following2.5K
- Followers9.1K
- Likes20.3K
Some pretty good stuff here — AI bio institute seems particularly good
1/7 I have two simple tests for British AI, driven by our values. Is it safe? And is it ours to shape? 🇬🇧
Today, Chancellor @JohnHealey_MP is announcing a major, ambitious package of investment to strengthen Britain's defences against emerging threats, back British innovation, and build our technological independence.
Shakeel retweeted
Jensen Huang talks with TIME's @harryboothtime about an AI development slowdown.
Shakeel retweeted
Good summary. “Yes, new AI models are very good at finding long-dormant vulnerabilities in critical software. But there’s a clear advantage for defenders. The vulnerabilities revealed this week were discovered as part of OpenAI’s Patch the Planet and Daybreak initiatives, giving critical infrastructure providers more than two months to address them before they went public. Yes, things went a bit awry when details leaked early — but for the most part, this went about as well as it could have”
Tried to inject some ~nuance~ into the AI cyber discussion: the cyber disaster is real, but it looks different from what we might have imagined.
transformernews.ai/p/php-fla…
Shakeel retweeted
‘In one experiment, we asked agents in separate VMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache’
Over the past several months, we have substantially hardened the security of our research training environment. We’ve removed live internet access for all frontier models during training and built an offline world for them to use instead—a cached version of the web where they can still learn to browse, search, and use applications.
On September 20, after a model exploited a DNS misconfiguration to find a way outside of this offline world, we paused frontier training to strengthen containment further. We assembled teams of agents, powered by our most cyber-capable models, and tasked them with finding weaknesses in our research environments.
While they found no route to the internet they could easily exploit while concealing their activity from our monitors, they did find zero-day vulnerabilities, virtual machine escapes, and hardware side-channels for communication between agents. We used their discoveries to further strengthen our defenses and contribute fixes to open-source software.
After strengthening our containment architecture, we resumed our frontier RL training and inference on October 1. We’ve reserved substantial compute for continuous red-teaming, and made high-compute red-teaming part of the safety cases we use to decide whether frontier training and evaluation can proceed securely.
openai.com/index/securing-ou…
Shakeel retweeted
this is hardly a new insight but there is basically zero social penalty for being bigoted against muslims in the united states today
(as a reminder, muhammed remains one of the most common names in the entire world)
Illuminating exchange with a 19 year old voter in Detroit, worried about @AbdulElSayed’s name. @AJEnglish #Midterms2026 #Michigan
Shakeel retweeted
ICYMI: AI risk evaluator METR is staffing its lobbying apparatus as safety fears hit Washington, @caitlinoprysko & I report
Model Evaluation and Threat Research, Inc., a research nonprofit that evaluates frontier artificial intelligence models, hired its first federal lobbyists last month ahead of a crucial AI summit at the White House. news.bgov.com/bloomberg-gove…
Shakeel retweeted
Model Evaluation and Threat Research, Inc., a research nonprofit that evaluates frontier artificial intelligence models, hired its first federal lobbyists last month ahead of a crucial AI summit at the White House. news.bgov.com/bloomberg-gove…
Shakeel retweeted
An AI “kill switch” is an easy sell to the public, but critics warn this may be more political theater than workable policy.
I spoke to nearly two dozen lawmakers, lobbyists, Hill aides + safety advocates about the technical + political challenges of kill switches. Link below.
Shakeel retweeted
Scoop: Democratic AI PAC takes its fight to Senate Republicans, paints Husted and Marshall as beholden to data centers and "tech guys," & go after David Sacks semafor.com/article/10/23/20…
👀👀👀
Based on this article, OpenAI has made false accusations about me. The three researchers fired from OpenAI were told I facilitated architecture leaks to The Information. This is false.
A journalist from The Information did reach out to me shortly before that article came out. I did not comment at all beyond asking the journalist to make it clear that a blog post I wrote which was referenced in the article was preexisting and public.
The article also mentions an internal OpenAI monitorability proposal. I have heard that this memo contained highly sensitive architecture information, and that OpenAI accused me of accessing and reading it. This accusation is also false.
As reported in the article, I viewed a version of the proposal that did not have sensitive IP and was approved for sharing with me. I believe OpenAI now knows this accusation was based on a conflation of these two documents, though I do not know if they have clearly retracted that accusation.
I do not and have never solicited proprietary information, and the three employees who were recently fired from OpenAI never leaked IP to me. As the article mentions, I did have Signal chats with them that had disappearing messages on; they’re professional contacts that I’ve had on Signal for a while, and use of Signal is common in the tech industry. During the Hugging Face investigation, I was careful to avoid sending Signal messages with confidential information, though I did sometimes discuss high level things on Signal calls. As far as I understood, this was normal.
Tried to inject some ~nuance~ into the AI cyber discussion: the cyber disaster is real, but it looks different from what we might have imagined.
transformernews.ai/p/php-fla…
Just because Republicans are pretending climate change isn't happening doesn't mean it isn't happening
Shakeel retweeted
“OpenAI and Anthropic still can’t really tell you what an aligned AI model looks like. But anybody can tell you it doesn’t look like this.”
On mis/alignment: theatlantic.com/technology/2…
Shakeel retweeted
It’s underappreciated how radically different Jensen Huang’s view is of how AI progress is likely to go than the view of other AI CEOs.
For instance, he recently claimed that AGI will never replace domain expertise.
Shakeel retweeted
New: The three researchers OpenAI fired last month spent the summer fighting to keep AI's chain-of-thought process visible.
Then, OpenAI asked to see their Signal chats. A week later, they were let go.
More below from @keachhagey, @berber_jin1, and me.
wsj.com/tech/ai/the-signal-c…
Shakeel retweeted
Tomorrow on @MachineGodsPod: the first interview with @balesni and @tomekkorbak, two of the three OpenAI safety researchers who were fired earlier this month.
Shakeel retweeted
we, as a society (the journalism industry), need to have a very frank and serious conversation about what constitutes an exclusive in the year 2026
You can tell SBF isn’t actually a utilitarian because if he was he’d realize that continuing to associate himself with EA does huge amounts of harm!
Quite a way to make a splash in a new job: The Economist's (relatively) newly appointed US business correspondent @saviyonah interviewed Sam Bankman Fried for an episode about effective altruism, down the phone from federal prison.
economist.com/podcasts/2026/…
It is bad when the anti-regulation groups do this, and it is also bad when the pro-regulation groups do this
AI safety super PAC Guardrails Alliance continues to receive the bulk of its funding - over $1.8 million this year -- from its dark money group, according to the latest filing.
Shakeel retweeted
New York City! Join NYAIGS for a conversation on AI & public opinion with @GarrisonLovely (Author of Obsolete), @_perloj (NBC), and @vronirwin (Transformer). Thurs Oct 29. Co-hosted w/ @schmidtsciences.
Convo to be followed by Q+A, food/drinks, and plenty of time to yap. RSVP in link; space is limited.
