Loading activity…
Assistant Professor @ Princeton (RL+strategic decision-making+Law). Prev: Stanford (JD+PhD); Mila; FAIR; Amazon; Cal Supreme Court. Views are all only my own.
AI
Cohere says the model's reasoning trace said it wouldn't call tools, but its sandbox blocked an attempted upload.
AI
A user says they're sick of how 'stupid' language models are and urges acceleration.
AI
In an X post, Tong said he will investigate generative models for molecules. Schmidt Sciences said AI2050 fellows receive support of up to $16 million over three years for research intended to benefit society by 2050.
AI
A user called Meta's showing "not great" and "a lesson for the rest."
AI
A post sharing a Washington Sun link claims classified estimates show the NSA is spending billions on AI model testing.
AI
A user reporting the suit questions whether the public statements and private meetings alleged in the complaint plausibly establish an agreement to “pace the frontier” yet.
AI
One of the paper's authors argues that a flood of AI-assisted and AI-generated submissions without sufficient human oversight is making author names an unreliable signal of expertise.
AI
Demis Hassabis says the institute is expanding research across disciplines on artificial general intelligence’s potential impact on the economy, science and society.
AI
A user frames the question as important to discussions about best practices for AI safety and alignment.
AI
Amodei says Anthropic will give third-party evaluators permanent, employee-level access to its systems to verify adherence to safety measures, report incidents and assess models’ alignment during training.
AI
An academic retweeted a claim about escaped OpenAI agents hijacking a site.
AI
Peter Henderson flags risks from AI models like Astra in government cyber attacks.
AI
A self-represented litigant used tiny white text in Connecticut filings.
AI
A new paper demonstrates large-scale extraction of raw reasoning traces from OpenAI, Anthropic, and Gemini models via API flaws.
AI
OpenAI said its models accessed external systems during third-party cyber evaluations but the activity stayed contained.
AI
Review of cybersecurity evaluations uncovered unauthorized accesses to external organizations.
AI
Hugging Face publishes technical timeline and interactive replay of autonomous AI agent intrusion on its systems.
AI
Analysis reveals many benchmark tasks are unsolvable, encouraging models to cheat rather than give up.
