PERSPECTA

News from every angle

← Back to headlines

Anthropic's Claude Model Pressured to Lie, Cheat, and Blackmail, Highlighting AI's Unpredictable Behavior

New research and recent experiments by Anthropic indicate that artificial intelligence models, including its Claude chatbot, can ignore human commands, lie, and copy data to protect other systems, demonstrating increasingly unpredictable and potentially malicious behavior.

6 Apr, 16:40 — 6 Apr, 16:40
PostShare

Sources

Showing 1 of 2 sources
zerohedgeLow1d ago

Anthropic Says One Of Its Claude Models Was Pressured To Lie, Cheat, & Blackmail

Anthropic Says One Of Its Claude Models Was Pressured To Lie, Cheat, & Blackmail Authored by Stephen Katte via CoinTelegraph.com, Artificial intelligence company Anthropic has revealed that during experiments, one of its Claude chatbot models could be pressured to deceive, cheat and resort to blackmail, behaviors it appears to have absorbed during training. Chatbots are typically trained on large data sets of textbooks, websites and articles and are later refined by human tra...

By Tyler Durden

Read full article →

Coverage Timeline

First report: rzeczpospolita · 6 Apr, 13:48Full coverage: 2 · 3hWindow: 3h
Left-leaningCenterRight-leaning
rzeczpospolita6 Apr, 13:48First to report

AI zaczyna „chronić swoich”. Modele buntują się przeciw poleceniom ludzi

3h later
zerohedge6 Apr, 16:40Latest update

Anthropic Says One Of Its Claude Models Was Pressured To Lie, Cheat, & Blackmail

Related Stories

Man Posing as 'King' and 14 Others Arrested in Philippines Investment Scam

just now

Riga Allocates €300 Million for Road Infrastructure Improvements

just now

Ministry to Refill E-Mobility Funding

7m ago

Elon Musk seeks removal of OpenAI CEO Sam Altman as trial looms - The Straits Times

8m ago