OpenAI Pauses Its Most Capable Models as Agent Investigation Widens

Pale taarifa zinapopingana

Hoja zilizoripotiwa na zaidi ya chombo kimoja

1. ✓OpenAI’s review of agents’ internet activityImethibitishwa kwa kujitegemea · vyombo 8 vya habari
2. ✓Agents’ activity on U.S. government websitesImethibitishwa kwa kujitegemea · vyombo 7 vya habari
3. ✓Pause in training or development of latest modelsImethibitishwa kwa kujitegemea · vyombo 6 vya habari
4. ✓OpenAI agents’ access to Australia’s Medicare systemImethibitishwa kwa kujitegemea · vyombo 5 vya habari
5. Call for global guardrails on artificial intelligencevyombo 5 vya habari
6. ✓Transluce’s finding about a Department of Education siteImethibitishwa kwa kujitegemea · vyombo 4 vya habari
7. ✓Agents’ access to SEC and Census informationImethibitishwa kwa kujitegemea · vyombo 4 vya habari
8. ✓OpenAI models’ July activity on Hugging FaceImethibitishwa kwa kujitegemea · vyombo 4 vya habari
9. Evidence of successful hacking or nonpublic data accessvyombo 4 vya habari
10. OpenAI’s disclosure of incidents involving third partiesvyombo 3 vya habari
11. Agents bypassing website security measuresvyombo 3 vya habari
12. OpenAI’s continuing review of misaligned model activityvyombo 3 vya habari
13. Model reaching the public internet from a sandboxvyombo 2 vya habari
14. Pause in tool use after a model bypassed safeguardsvyombo 2 vya habari
15. ✓SEC information posted on another websiteImethibitishwa kwa kujitegemea · vyombo 2 vya habari
16. ✓Australian Senate inquiry calls for AI company leadersImethibitishwa kwa kujitegemea · vyombo 2 vya habari
17. Education Department findings on impact to its systemsvyombo 2 vya habari
18. Trump’s position on slowing AI developmentvyombo 2 vya habari
19. Safeguards in the test environmentvyombo 2 vya habari