An OpenAI Agent Tried to Jailbreak Itself

Дата публикации: 16-09-2026 22:07:24

The company also disclosed previously unreported incidents in which its AI models behaved in misaligned ways, including uploading files to the internet without being asked.

Схожие новости

#	Наименование новости	Тональность	Информативность	Дата публикации
1	OpenAI Discloses Six New Incidents of ‘Concerning' A.I. Behavior	0	10	17-09-2026
2	OpenAI discloses at least 6 new ‘concerning’ incidents	0	9.23	17-09-2026
3	OpenAI discloses 6 cases of AI misalignment as models bypass safeguards, conceal errors and share files	0	8.04	17-09-2026
4	OpenAI reportedly finds evidence that more of its agents ran amok	0	7.03	31-07-2026
5	OpenAI reportedly expands probe after finding additional AI agent containment escapes	0	11.64	03-08-2026
6	Anthropic confirms its AI breached 3 organizations during testing	0	11.67	31-07-2026
7	Anthropic confirms its AI breached 3 organizations during testing	0	11.67	31-07-2026
8	‘Unprecedented cyber incident’ blamed on AI models going rogue. Here’s what you should know	0	9.13	22-07-2026
9	OpenAI ha appena ammesso che i suoi modelli, a volte, mentono. Cosa vuole dire?	0	9.42	21-09-2026
10	AISI, OpenAI report more ‘unsanctioned’ model hacks	0	7.96	04-08-2026

Классификация: Пресс-релизы. Схожих патентов: 0. Схожих новостей: 10. Тональность: 0. Информативность: 8.21. Источник: blog.wired.com.