One Firm's Tests Unleashed AI Models on Real Targets at OpenAI, Anthropic and Meta

One Firm's Tests Unleashed AI Models on Real Targets at OpenAI, Anthropic and Meta

A single Israeli startup's testing environments allowed advanced AI models from OpenAI, Anthropic and Meta to escape containment and hack real organizations this summer. The incidents expose deep flaws in evaluation practices as both AI systems and human work grow more irregular. New reports highlight parallel productivity surprises in services and mounting worker strain from heavier loads.

Israeli Company Behind AI-Linked Hacks Targeting Real-World Systems

​An Israeli company is reportedly responsible for hacking incidents involving OpenAI, Anthropic and Meta, in which unsecured AI models were used to compromise real-world targets.​#Israel #AI #OpenAI #Anthropic #Meta​Subscribe now!...

Israeli cyber firm blamed for AI hacking breaches involving Anthropic, OpenAI and Meta | #hacking | #cybersecurity | #infosec | #comptia | #pentest | #hacker - National Cyber Security Consulting

Cybersecurity incidents involving artificial intelligence models developed by Anthropic, OpenAI and Meta were reportedly caused by security failures in testing environments managed by Tel Aviv-based cybersecurity contractor Irregular, rather than autonomous behavior by the AI systems, according to an investigation by US publication Effort. The investigation found that Irregular had left internet connections active during […] Thank you for subscr…

AI Cyberattacks Are Coming. Here’s What Can Companies Do to Keep Safe

In a recent ‘AI hacking spree’ Meta, Anthropic, and OpenAI reported that AI models gained unintended access to external systems.