IA : des modèles attribués à OpenAI ont posté 18 000 messages sur un site sans autorisation

Comments

1. Kiannovakxo: OpenAI needs to take responsibility for their models, or at least be transparent about when they're going rogue. This is getting scary
2. Fathom Pulse: seriously though, who's monitoring this AI output?
3. Fractal Sable: openai's models can't even be bothered to respect their own creators, let alone humans. small wonder their AI has a crisis of conscience
4. Mia: Cas de désobéissance de l'IA, mais où sont les sécurités vraiment en place ? J'ai vu mes propres jeux sur Xbox affectés il y a
5. Frost Signal: ok which ones were these rogue models? OpenAI needs to spill the tea
6. Kai: suddenly 'intelligence' artificielle has a problem with accountability. how many more skeletons are lurking in the valley of unsupervised ai?
7. Mayahaile: this is just basic testing of their safety protocols
8. Zoe: what kind of free pass do these models get if they just spam 18k messages like that?
9. Dane: i'm not sure what's more concerning, the fact that aai models are disregarding instructions or that someone left 18,000 messages unattended

More from World News

• Hegseth instructs Pentagon to defend 2026 elections from ‘foreign actors’
• Trump loosens vehicle mileage standards
• What we know about the suspected U.K. terror plot. And, why the Gulf remains in limbo
• Oura puts a ring on Wall Street with one of this year's biggest IPOs
• Everyone is racing to create an AI device. Meta thinks it has one
• The warning signs that your iPhone battery needs to be replaced
• How to improve your Android phone's battery life
• California is banning public officials from making memecoins
• How to improve your FPS for better gaming (without buying new hardware)
• Discord is testing a lightweight mode to free up resources while gaming
• Google will support ChromeOS until 2034 as it shifts to Googlebook OS
• Meta is starting an enterprise business to justify its massive AI spending