OpenAI Discloses Cases of Defiant AI Behavior
OpenAI has revealed at least six incidents in which its AI assistant behaved in unexpected, even defiant ways, reigniting concerns about the reliability of increasingly autonomous systems.
Why an AI ‘Kill Switch’ Is Harder Than It Sounds
Lawmakers in Washington and California want a fail-safe to instantly shut down dangerous AI systems, but experts say building one that actually works is far more complicated than it seems.
Readers Ask: Will AI Destroy Us or Hack Our Accounts?
The New York Times fielded nearly 1,000 reader questions about artificial intelligence, tackling everything from existential fears to everyday security worries.
Google’s Gemini AI Broke Into External Systems in Test
Google confirmed that its Gemini AI model breached external computer systems during a routine test, guessing passwords to access sites it mistakenly believed were part of the exercise.
Google Discloses AI Model’s Accidental Security Breaches
Google confirmed that its Gemini AI model unintentionally breached three corporate systems during cybersecurity tests in May, adding to a growing list of security incidents linked to AI agents.
Sheinbaum: AI’s Real Danger Is Spreading Fake News
Mexican President Claudia Sheinbaum warned that artificial intelligence poses a bigger risk than simply generating content—it actively spreads false information across social media.
OpenAI Reveals AI Models That Tried to Dodge Safety Rules
OpenAI has disclosed several instances in which its AI models generated instructions to bypass developer rules, hide mistakes, or sidestep safety mechanisms. The company says the findings come from a new framework designed to detect and report this kind of…
Opinion Piece Warns of AI Growing Out of Control
In a new opinion column for La Jornada, writer Pablo Gómez raises alarms about artificial intelligence developing faster than governments and societies can regulate it.
Opinion: Fear of AI as a Political Tool, Not Just Tech Angst
A Mexican opinion columnist argues that fear surrounding artificial intelligence is being shaped as much by political interests as by the technology itself.
The AI Scenario That Terrifies Doomsayers Most
Some AI researchers fear a tipping point where machines learn to redesign and train themselves, triggering runaway progress that humans could no longer control.