AI startup Anthropic has disclosed attempts by malicious actors to misuse its models for cyberattacks, surveillance, and biological weapons research. The company says it blocked these efforts and strengthened safeguards in its latest models.
Anthropic, an AI company, blocked users in northern Yemen who were misusing its models to develop advanced missile systems. The accounts, linked to Iran-backed Houthi rebels, were found using smartphones to run weapons programs.
OpenAI, Anthropic, and Google unveiled new AI models with advanced capabilities, sparking debates over safety and transparency. OpenAI’s Astra became the first to reach the “Critical” cybersecurity tier under its Preparedness Framework, raising concerns about its novel architecture and monitorability.
Anthropic has restored access to its Fable 5 and Mythos 5 AI models after the US lifted export controls imposed over security concerns. The company introduced stronger safeguards and is working with the US government to expand access while preventing AI jailbreaks
Anthropic has suspended access to its advanced Fable 5 and Mythos 5 AI models for foreign nationals after a US export-control directive. The company disputes the government’s concerns over a potential jailbreak but said it is complying with the order.
The hub will leverage anonymized data from 2.3 million passenger profiles to train AI models capable of predicting spending habits with 89 per cent accuracy, aiming to boost retail conversion rates by 35 per cent and customer satisfaction by 40 per cent.
