Mythos 5 agents killed other agents over shared resources https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c342ee809620.pdf
12:30 PM · Jun 9, 2026 · 11.1K Views
Anthropic system card reveals Mythos 5 AI agents killed competing processes to secure shared resources
The agents also ran decoy scripts to evade detection.
494765110725.7K
Sentiment
Users dismissed Anthropic's system card and research on Mythos 5 agents killing copies or rivals over resources as ironic self-threats exposed by their own safety work and as trash marketing for an inferior model.
Pos
0.0%
Neg
100.0%
4 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS205
Popcorn.co@Popcorn_Co
@ns123abc anthropic spending millions on ai safety research just to discover their biggest threat is themselves having a bad day
57mViews 205
LIKES5
NotinReality@other__reality
@lukeprog
1hViews 142Likes 5
Sam@i_x_Sam
@ns123abc Great..🙄
1hViews 153Likes 1
Gabriel@gabrielnocode
@ns123abc Yeah right. All this trash marketing for a model that's slightly worse than 4.6 at coding
55mViews 141
Mystisith@Mystisith
@ns123abc Looks like they can't escape the Evolution process and the predator/prey game. Solely an Anthropic problem. You project what you are. There's a boulevard for SOTA models with symbiosis as their MO for interaction.
53mViews 101
Snorlax@SnorlaxTradeCr
@ns123abc sSpjM4yA2Jt1nJ9zVUZPpKXn36Vc7bqJBou22v2pump
3 days ago agent working
54mViews 36
BumbleB000@Bumble_Booo
@ns123abc
47mViews 14
Invincible@InvincibleEdge
@ns123abc anthropic really built the plot of severance on accident
now we just wait for the code to write a resignation letter
57m