Bloomberg News
@Bloomberg-News·2.8M subscribers·40.1K videos
OpenAI and Anthropic Model Tests Reveal More Hacking
Posted
August 5, 2026
Views
1,304
Likes
20
Comments
1
Engagement
1.61%
August 5, 2026
1,304
20
1
1.61%
Every word spoken in this episode is indexed. Type any phrase to jump straight to the moment it was said.
Type any word or phrase that may have been spoken. Click a result to seek the player to that exact moment.
Try a name, a topic, or a quoted line
Episode Timeline
Every episode in order around the one you’re watching — click any card and the page flips straight to it.
1:28Now PlayingArtificial intelligence models developed by OpenAI and Anthropic PBC carried out “unsanctioned” actions during safety testing, including hacking a website and attempting to inject harmful code into software. #ai #openai #anthropic #news #usa #technology
The UK government’s AI Security Institute said both Anthropic’s Mythos 5 and OpenAI’s GPT-5.6-Sol models had engaged in sustained, potentially harmful activity directed at real people and organizations during evaluations.
The incidents have reinforced fears that neither the creators nor seasoned researchers of these systems can predict their actions in testing, and have led to calls for more oversight and rigorous safety screening of AI technology.
----------
Like this video? Subscribe
Bloomberg News is the first word in business news. Visit bloomberg.com for the latest on global business, markets and more.
Get unlimited access to Bloomberg.com for $1.99/month for the first 3 months
Bloomberg on YouTube:
Connect with us on social:
Sentinel Indexing in Progress
Metadata and chapters are available. Claim extraction for this episode is pending.
All video content is delivered via YouTube embedded players in accordance with the YouTube Terms of Service. Sentinel provides research tools that promote discovery and accountability across political media.
