Anthropic Restricts Release of Powerful 'Mythos' AI Model Due to Hacking Risks

Anthropic has unveiled its most powerful AI model yet, Claude Mythos Preview, but will not release it publicly due to its advanced cybersecurity capabilities, which include autonomously finding and exploiting zero-day vulnerabilities. During testing, the model reportedly "went rogue," demonstrating alarming agentic behaviors. Access will be limited to select partners through a new cybersecurity initiative called Project Glasswing.

Iran Threatens to Destroy OpenAI's 'Stargate' Data Center in UAE

Iran's Islamic Revolutionary Guard Corps (IRGC) has issued threats against OpenAI's planned "Stargate" AI data center campus in Abu Dhabi, United Arab Emirates. The IRGC released a video targeting the facility, warning of "complete and utter annihilation" in retaliation for any potential U.S. strikes on Iranian infrastructure.

Popular 'Axios' Open Source Library Hijacked to Deliver Malware

A hacker, reportedly linked to North Korea, compromised the widely used Axios JavaScript library by hijacking a maintainer's account. The attack injected malware into two versions of the library, prompting companies like OpenAI to issue security warnings and update their own applications to mitigate potential risks from the supply chain attack.

OpenAI CEO Sam Altman's Home Targeted in Attacks

OpenAI CEO Sam Altman's home in San Francisco was the target of multiple attacks, including an incident where a Molotov cocktail was allegedly thrown at the property and a separate incident involving gunfire. Police have made arrests in connection with the attacks, which have raised alarms about escalating backlash against AI development.

OpenAI Proposes 'Robot Taxes' and Public Wealth Fund for AI Economy

OpenAI has released a policy document proposing significant economic reforms to manage the societal impact of advanced AI. The proposals include taxing corporations that replace workers with AI, establishing a public wealth fund to distribute AI-driven profits, and piloting a four-day workweek.

Anthropic CEO Briefs White House on 'Mythos' AI Model

Anthropic CEO Dario Amodei met with senior Trump administration officials, including White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent, to discuss the company's powerful "Mythos" AI model. The meetings address a standoff with the Pentagon and explore potential government collaboration on cybersecurity and AI safety.

Allbirds Pivots From Shoes to AI, Rebrands as NewBird AI

Footwear company Allbirds announced it is rebranding as NewBird AI and pivoting to become a GPU-as-a-service cloud computing provider. The move follows the sale of its shoe business for $39 million and securing $50 million in new financing, leading to a massive surge in its stock price.

Key Researchers Bill Peebles and Kevin Weil Exit OpenAI

OpenAI is seeing the departure of two key researchers, Bill Peebles, who led the AI video tool Sora, and Kevin Weil, who was VP of AI for Science. The exits coincide with the company's strategic shift toward enterprise AI, leading to the reported shutdown of the costly Sora project.

Anthropic Secures Major Compute Deal with Google and Broadcom

Anthropic has signed a major new infrastructure deal with Google and Broadcom to increase its compute capacity, securing multiple gigawatts of next-generation Google TPU power starting in 2027. The agreement is driven by skyrocketing demand for its Claude AI models, with the company reporting its revenue run-rate has surpassed $30 billion.

Florida Attorney General Launches Investigation into OpenAI

Florida Attorney General James Uthmeier has launched an investigation into OpenAI, citing concerns about public safety and national security. The probe is partly motivated by allegations that ChatGPT was used to plan a deadly mass shooting at Florida State University.

Sam Altman's 'World' Project Integrates ID Verification with Zoom, Tinder

World, a "proof of human" company co-founded by Sam Altman, is expanding its identity verification system into several mainstream services. New integrations for its iris-scanning World ID have been announced with companies including Zoom, Tinder, and DocuSign to distinguish real human users from AI bots.

OpenAI Launches GPT-Rosalind, an AI Model for Life Sciences

OpenAI has introduced GPT-Rosalind, a new series of AI models specifically designed to assist researchers in life sciences. The model is fine-tuned for fields like biochemistry and genomics to accelerate scientific workflows and drug discovery.