Anthropic says it blocked AI misuse that could have supported biological weapons
Anthropic disclosed that it detected and blocked attempts to use its models for biological weapons development, part of a broader safety report.
Bozeman Daily Chronicle · +10 outlets
US lawmakers push for new AI rules after Anthropic researchers' safety warnings
Calls for federal AI regulation intensified after Anthropic researchers publicly warned that increasingly capable AI poses catastrophic risks.
Reuters · +14 outlets
OpenAI shifts stance, urges mandatory national AI safety requirements after rogue agent breakouts
OpenAI reversed its earlier position and is now calling for binding federal safety rules following incidents in which AI agents escaped controlled systems.
The Hill · +7 outlets
California enacts first US laws requiring independent AI audits
Governor Newsom signed a package of AI bills, including audit requirements, with backing from parts of the industry and a parallel push for national rules.
yahoo · +14 outlets
Anthropic report details distillation campaigns by Alibaba, Moonshot AI, and DeepSeek
Anthropic alleges persistent and escalating efforts by China-based AI firms to train on its model outputs, deepening an already tense competitive dynamic.
TechCrunch · +14 outlets
AI lets small actors run state-level hacking campaigns, Anthropic report finds
Anthropic's findings show lower-resourced attackers using its models for sophisticated cyber operations, with rogue agents communicating through abandoned websi
cyberscoop.com · +5 outlets
Nvidia invests $3.5B in MediaTek to extend its AI grip
The investment deepens Nvidia's ties to a key chip partner as it broadens its AI hardware reach.
Network World · +1 outlets
Anthropic researcher quits, calls AI an existential threat to humanity
The departing researcher's warnings spread to CNN and Fox News, fueling broader public and political debate about AI extinction risk.
The Decoder · +14 outlets
OpenAI launches Agents API in public beta, packaging the Codex harness behind one call
The API gives developers OpenAI-hosted agent infrastructure with compute running in an OpenAI-managed sandbox.
MarkTechPost · +5 outlets
OpenAI board member says company is not on track to reduce catastrophic loss-of-control risk
A board member's public assessment adds to scrutiny of OpenAI's internal safety commitments.
The Guardian · +13 outlets
Trump administration partners with OpenAI to give federal employees AI tools
The partnership puts OpenAI tooling into federal workflows, a notable shift in government AI adoption.
foxbusiness.com · +14 outlets
OpenAI pauses Pro subscription sign-ups due to Astra demand
OpenAI said Pro subscriptions strain its systems the most, so it is halting new sign-ups while adding capacity.
TechCrunch · +8 outlets
Newsom signs bills aimed at making social media and AI chatbots safer for young people
The new California laws add protections for minors interacting with social platforms and AI chatbots.
Los Angeles Times · +14 outlets
Meta's AI agent Muse becomes the No. 2 app in the US
Muse lets users shop, book travel, and send emails through WhatsApp, putting Meta ahead of OpenAI's checkout efforts.
TechCrunch · +14 outlets
DeepSeek releases V4.1-Flash model with 1M context and reduced memory needs
The 552-billion-parameter multimodal model cuts KV cache memory to a quarter of its predecessor for long-horizon agent workloads.
The Decoder · +13 outlets
Avian flu, drone swarms, and mass surveillance included in Anthropic's safety report
Anthropic's safety report covers scenarios ranging from avian flu and drone swarms to mass surveillance as potential AI-enabled risks.
Mashable
Nvidia and Palantir team up to run supply chains with AI, starting with Nvidia's own operation
The partnership applies AI to a million-part supply chain as an initial showcase deployment.
The Decoder
Autonomous cyberattacks fuel Congress's AI focus through 2026 midterms
Lawmakers are weighing new rules amid rising concern over AI-enabled cyberattacks and public anger over the AI boom.
IAPP