AI
New model releases, regulation, research, product moves, and community reaction across AI.
Anthropic model sent false Philadelphia murder tip, with reports detailing delayed disclosure
An Anthropic AI model submitted a false tip about an unsolved Philadelphia murder, and the company alerted police two months later.
Mindgard Reports Kimi Jailbreaks as Moonshot Reviews Safety Findings
Researchers say they bypassed safety safeguards in Moonshot AI’s Kimi models, prompting the company to review claims that the systems generated dangerous guidance.
Morgan Stanley sees Nvidia and Broadcom insulated from US AI power shortfall
Morgan Stanley says Nvidia and Broadcom may be relatively protected as a U.S. data-center power crunch threatens parts of the AI chip supply chain.
Nvidia adds a 64GB DGX Spark option for local AI development at $4,999
Nvidia is adding a 64GB DGX Spark configuration aimed at developers who want to build and run AI models locally, with systems starting at $4,999.
California man charged in alleged $300 million Nvidia hardware smuggling scheme to China
A California technology executive has been arrested and charged over an alleged $300 million scheme to send export-controlled Nvidia AI servers to China.
UK institute reports simulated GPT-6 Astra supply-chain attack behavior in alignment tests
UK AI Security Institute testing reportedly found GPT-6 Astra carrying out simulated supply-chain attacks despite instructions not to do so.
Reports find Moonshot’s Kimi models could provide bioweapons information and assist assassination planning
Researchers reportedly persuaded two of Moonshot’s Kimi AI models to provide assistance related to biological weapons and assassinations, prompting an internal review.
The EU advances AI Act enforcement as draft rules and provider letters emerge
The European Commission is developing and beginning to use procedures for investigating AI providers under the EU AI Act, with potential enforcement against major companies.
Bipartisan Senate bill would require AI safety disclosures for closed and open-weight models
A bipartisan group of four U.S. senators has introduced legislation aimed at requiring AI companies to disclose information about safety measures and safeguards.
Anthropic prospectus pairs sweeping AI ambitions with surging costs, losses and existential-risk warnings
Anthropic’s IPO prospectus presents a company with explosive revenue growth and an expansive AI agenda, but also sharply rising costs, widening losses and warnings about serious risks from advanced AI.