Breaking: Autonomous Agents are a Shitshow
We recorded 4 claims from this post. 3 were checked against evidence. 1 can’t be checked. 1 more is held. Read the post at its source.
Claims from this post (4)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: Can’t be checkedSaid
Rhetoric
“We were not wrong.”
Our reading: Marcus and Hamiel's earlier warning that LLM coding agents are a security nightmare has been borne out.
No horizon
Qualifier: refers to August essay 'LLM + Coding Agents = Security Nightmare'
Assessment: SupportedSaid
Observation
“A new study from researchers from Stanford, MIT CSAIL, Carnegie Mellon, ITU Copenhagen, and NVIDIA and Elloe AI Labs, examining 847 autonomous agent deployments drawn from healthcare, finance, customer service and code-generation, showed that 91% were vulnerable to subtle but dangerous tool-chaining attacks (see diagram below),”
Our reading: A new study found 91% of 847 autonomous agent deployments were vulnerable to tool-chaining attacks.
No horizon Evidence as of
Limitation: The primary paper was not found online. The assessment rests on news reports.
Assessment: SupportedSaid
Observation
“Additionally, in the new study 89.4% of agents showed drift relative to their goals after about 30 steps in their process, and 94% of agents with some form of memory-augmentation were vulnerable to poisoning attacks.”
Our reading: In the new study, 89.4% of agents drifted from goals after about 30 steps, and 94% of memory-augmented agents were vulnerable to poisoning.
No horizon Evidence as of
Limitation: The primary paper was not found. The assessment rests on news reports.
Assessment: SupportedSaid
Observation
“The new paper very much amplifies what a team of AWS and Berkeley researchers showed in February, documenting similar vulnerabilities:”
Our reading: The new findings on agent vulnerabilities reinforce similar vulnerabilities documented by AWS and Berkeley researchers in February 2026.
No horizon Evidence as of
Limitation: Marcus does not name the February paper; it is identified as STAC by inference.
Held from this post (1)
These records did not pass every check, so nothing about them is published. Held is not a finding.