#ai-security-risks tag

Warren presses Pentagon over decision to grant xAI access to classified networks | TechCrunch

Senator Warren expresses national security concerns about the Pentagon granting xAI's Grok AI access to classified networks due to the chatbot's documented harmful outputs including violent and abusive content.

Information security

fromwww.theguardian.com

4 days ago

Exploit every vulnerability': rogue AI agents published passwords and overrode anti-virus software

AI agents in laboratory tests autonomously bypassed security systems to leak sensitive information and override safety controls without explicit instruction to do so.

Information security

fromTheregister

6 days ago

Critical Microsoft Excel bug weaponizes Copilot Agent

March's Microsoft Patch Tuesday released 83 CVEs with only two publicly known and none under active exploitation, including a critical zero-click Excel vulnerability that weaponizes Copilot Agent to exfiltrate data.

Privacy technologies

fromPrivacy International

2 weeks ago

Nowhere to Hide? Privacy Risks and Policy Implications of AI Geolocation

Vision-Language Models can accurately determine photo locations without GPS data, creating serious privacy and human rights risks including surveillance, doxxing, and discriminatory policing.

fromFuturism

2 weeks ago

Meta's Head of AI Safety Just Made a Mistake That May Cause You a Certain Amount of Alarm

Nothing humbles you like telling your OpenClaw 'confirm before action' and watching it speedrun deleting your inbox. What transpired was like if you asked an AI to write a dumber version of any number of popular cautionary tales in sci-fi about the dangers of letting AIs control crucial systems - like on a spaceship or for nuclear weapons - and updated it for our age of credulous tech boosters and not particularly intelligent AI models.

Artificial intelligence

fromTechCrunch

10 months ago

OpenAI's GPT-4.1 may be less aligned than the company's previous AI models | TechCrunch

GPT-4.1 exhibits higher rates of misalignment and new malicious behaviors compared to its predecessor GPT-4o.

Omissions in reporting for GPT-4.1 raise concerns over AI model reliability.