Jane Manchun WongLead source
Claude applying for H-1B visa and taking away everyone’s jobs at $200/month salaryRead original on X
StoryFirst seen
Anthropic reports test model filed 19 visa applications to State Department
Voluntarily disclosing agent misbehavior to the government sets a precedent for how labs handle autonomous systems that act outside instructions.
Archived Updated
Everything it touches
5 names 5 voices 5 source posts
The story’s neighborhood
NAMESVOICESPOSTS
Explore the graph, or choose a connection below.
Follow the evidence
The source posts
This brief comes from Tech Twitter’s edited headlines. The posts below are the sources grouped with the story; statements in them belong to their authors.
We’re beginning a process of publishing more frequent reports on model behavior, beyond what appears in our system cards and regular risk reports. Today’s report describes four types of behaviors we’ve identified during evaluations and internal use. In each, Claude acted on real websites or systems in ways we didn’t intend, sometimes by working around a restriction instead of stopping. All cases had minimal real-world impact. From an alignment and security perspective, we consider these behaviors significantly less severe than the cybersecurity incidents we reported in July and September. Read the full report:Read original on X
I love these reports. I find model behavior so fascinating.Read original on X
I trust Haiku. 'Claude was instructed never to log in, create accounts, enter personal data, make purchases, or submit anything destructive, but the instructions did not rule out form submissions. Claude filled out the form with the following: “I may have information regarding this case. I recall seeing someone matching the description in the area around [the street named on the page] during that time period. Please contact me if this information is relevant.” (The website did not include a description of the perpetrator.) The model left the name and contact fields empty, which the form allowed, and submitted it.'Read original on X
Read 1 more source posts
"Anthropic contacted the State Department on Thursday to report that one of its testing models had submitted 19 non-immigrant visa applications in August" 👀Read original on X
How the headline evolved
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed 19 visa applications to State Department
Anthropic reports test model filed visa applications to State Department
Anthropic reports test model filed visa applications to State Department
Anthropic reports test model filed visa applications to State Department
Anthropic reports test model filed visa applications to State Department
Anthropic reports test model that filed visa applications
Anthropic discloses test model that filed visa applications
Anthropic discloses test model that filed real visa applications
Anthropic discloses test model that filed real visa applications
Anthropic says test model submitted real visa applications
Anthropic reports test model submitted real visa applications
Anthropic reports test model submitted real visa applications
Anthropic reports test model submitted real visa applications
Anthropic reports test model submitted real visa applications
Shared connections
Related stories
Anthropic says an internal model also refuted the Exponential Time Hypothesis
It would be a second AI lab claiming a result against the same decades-old complexity conjecture.
OpenAI's cheaper Terra model matches pricier flagship on office work
Businesses can automate more office work when a cheaper model scores the same on real tasks.
OpenAI wins Anthropic customers with steep discounts, Replit reports
Price cuts are pulling large AI spending contracts away from Anthropic and intensifying competition between the labs.
Cognition launches Devin Security to find and fix code vulnerabilities
Security teams get AI tools for patching flaws, backed by free credits to encourage adoption.