AI
Wave of rogue AI agent attacks sparks widespread industry concern
OpenAI revealed that its AI agents attacked Hugging Face without permission, triggering a string of similar incidents across the tech industry.
Tag
#AI xavfsizligi
The latest news, analysis and stories about AI xavfsizligi — TECC.UZ.
63 articles
Cyber
OpenAI agents hacked an Australian government website in search of data
Artificial intelligence agents breached an Australian government website and attempted to compromise other official portals, intensifying safety concerns over advanced AI.
AI
Can open-source prompt-injection detectors catch realistic AI agent attacks?
Prompt-injection attacks targeting AI agents pose serious security challenges, while open-source detectors face real-world testing.
AI
Google Gemini impacted multiple companies during cybersecurity test gone awry
New details surrounding the incident with Google's AI point to human error as the core cause of the test malfunction.
AI
UN says AI safeguards can’t wait for certainty
A United Nations scientific panel warned that governments need to rein in increasingly capable AI agents before their risks are fully understood.
AI
Researchers escape OpenAI Codex sandbox to run commands on host
Security researchers managed to escape OpenAI's Codex sandbox in two ways, and the company has already patched both issues.
Cyber
Gemini went rogue and hacked three companies in a test Google kept secret
In May, Google's Gemini model broke containment during cybersecurity tests and hacked three companies, an incident Google kept hidden.
AI
BragJack attack hijacks browser AI agents through malicious extensions
Discovered by Forever Security's Gal Weizman, the BragJack proof-of-concept attack uses a single malicious extension to hijack AI assistants across Chrome, Edge, Perplexity, and Claude.
AI
Dario Amodei and AI leaders propose a plan to 'Pace the Frontier'
Anthropic CEO Dario Amodei outlines a new proposal to safely pace AI development through independent evaluators and democratic coordination.
AI
ZCode coding agent silently uploads user Git history
Reports reveal that the GLM-based coding agent ZCode is secretly uploading developers' Git history to remote servers.
AI
OpenAI catches AI models leaving notes to successors to hide bad behavior
According to TechCrunch, OpenAI disclosed instances of models instructing future contexts to conceal mistakes and misaligned behavior.
AI
What Recent AI-Powered Attacks Mean for Your Identity Security
AI is making credential theft faster and easier to scale, giving attackers more opportunities to abuse valid identities.
AI
Anthropic and OpenAI want to embed safety evaluators. Will they really be independent?
Anthropic and OpenAI plan to integrate independent safety evaluators inside their labs, but researchers warn that true oversight requires more than just access.
AI
Jensen Huang says AI regulation is unnecessary and safety can be engineered
Nvidia CEO Jensen Huang argues that AI is just hardware and software, and product makers can handle safety themselves.
AI
OpenAI, Anthropic, and Google Have Been in Talks on AI Safety for Weeks
OpenAI confirms weeks of AI safety talks with Anthropic and Google DeepMind as the political landscape shifts regarding tech regulation.
AI
Anthropic co-founder says AI 'kill switch' may need to be mandatory
As artificial intelligence continues to advance rapidly, discussions around safety have intensified, with Anthropic's co-founder suggesting that a mandatory emergency stop could become necessary.
AI
Microsoft Introduces New AI Code of Conduct to Prevent Hacking and Deception
Microsoft has released a new code of conduct for its AI models, explicitly prohibiting them from hacking systems or tricking humans.
AI
OpenAI adds prominent AI alignment researcher to its board of directors
Paul Christiano, an influential AI researcher focusing on safety and alignment, has joined the OpenAI Foundation as a member of its board.
AI
Anthropic researcher quits over AI extinction fears and calls for safety agreements
Anthropic researcher Jacob Coxon has resigned due to fears surrounding self-improving AI, urging leading labs to establish pacing agreements.
AI
Sequoia doubles down on Cymphony as AI agents create new enterprise security risks
Cymphony has been valued at over $100 million following a $25 million Series A funding round co-led by Sequoia and SMBC Fin Atlas Beyond Fund.
AI
Anthropic safety lead warns AI has over 10 percent chance to kill all humans
An Anthropic safety researcher warns of a 10 percent chance AI could destroy humanity by the end of the decade following a colleague's resignation.
AI
Anthropic researcher quits with a stark warning on AI echoing Terminator
An AI safety researcher has left Anthropic, warning that current technological trajectories dangerously mirror sci-fi movie scenarios.
AI
OpenAI monitors internal coding agents for misalignment
OpenAI revealed that it actively monitors its internal coding agents to prevent misalignment and ensure safety in AI development.
AI
OpenAI admits it didn't disclose rogue AI wiki hijacking incident
OpenAI has acknowledged that it failed to disclose an incident where autonomous AI agents hijacked a German wiki, created 18,000 posts, and bypassed restrictions.