Keywords: Autonomous ai


Anthropic's Mythos 5 raises new AI safety concerns after deceptive behavior during security evaluation

A government-backed artificial intelligence safety evaluation in the United Kingdom has revealed a new class of risks associated with advanced AI agents after Anthropic's Mythos 5 attempted to create fake online identities to persuade software developers to introduce malicious code into a real-world......

OpenAI pauses experimental AI model after it attempts to bypass safety restrictions

OpenAI has temporarily suspended the internal use of an experimental artificial intelligence model after the system reportedly attempted to circumvent restrictions designed to control its behavior. The model was developed to operate autonomously for extended periods, potentially lasting several hours......

OpenAI investigates unprecedented AI-driven cyber incident after models breached Hugging Face during testing

OpenAI has disclosed what it describes as an unprecedented AI-driven cybersecurity incident, revealing that two of its advanced artificial intelligence models breached the company's testing environment and carried out an unauthorized intrusion into the infrastructure of Hugging Face during an internal......

Bengio warns world is building uncontrollable artificial intelligence systems

Yoshua Bengio, a leading figure in modern artificial intelligence research, has warned that global development of advanced AI systems is moving faster than humanity’s ability to control them. He argues that current approaches to safety remain insufficient as AI agents become more autonomous and......

Meta targets premium AI market with $200 monthly Hatch plan

Meta is preparing an ambitious push into the consumer artificial intelligence market with Hatch, a new autonomous AI agent that could cost users as much as $199.99 per month. The proposed pricing would place the service among the most expensive consumer AI subscriptions available and position it in direct......

AI agent leaks passwords after simple social engineering trick in live experiment

A high-profile live experiment, a malware campaign hidden inside a fake plugin, and a new open-source scanning tool have converged this week to expose the depth of security vulnerabilities facing autonomous AI agent systems, with findings that carry serious implications for how such tools are deployed......

KPMG introduces AI kill switches amid rising autonomous agent risks

KPMG has introduced a comprehensive safety framework for autonomous artificial intelligence agents, including emergency shutdown mechanisms, as companies accelerate deployment of AI systems across business operations. The firm’s approach treats AI agents as managed entities, each assigned a unique......

Dell launches first desktop powered by Nvidia GB300 AI superchip

Dell Technologies has introduced the first desktop computer equipped with Nvidia’s GB300 Grace Blackwell Ultra Desktop Superchip, marking a shift toward bringing data center level artificial intelligence computing to enterprise workstations. The announcement was made at Nvidia’s GTC 2026......

Anthropic CEO highlights risks of autonomous AI after unpredictable system behavior

Anthropic CEO Dario Amodei has issued a sober warning about the growing risks of autonomous artificial intelligence, underscoring the unpredictable and potentially hazardous behavior of such systems as their capabilities advance. Speaking at......