Signals
A number moved. Here is what it means for the person holding the job.
Signals is the daily desk. Labour-market data, capability benchmarks, filings, hearings, the quarterly numbers where a company admits what it is really doing with the technology. Each piece takes one finding, names the source it came from, and says in plain terms what changes for an operator — the person whose work, pay or position depends on it.
What a signal is not: a product launch, a funding round, a demo. Those are noise until they change who gets hired, what gets paid for, or which task stops being done by a person. When they do, they become a signal and they appear here.
One a day, written from the primary source, with every figure checked against the page it came from before it publishes. Each ends with the move to make this week, not a forecast. Start with the newest; the topics above sort the archive by subject.
Signals · Public Opinion
Only one in five Americans expects AI to make people more productive, and that gap is where operators earn their standing
3 min read
Signals · Judgement
A $50 site of invented statistics reached a bank's research. Checking the number is now the job
4 min read
Signals · Hiring
Anthropic's junior share of technical hires fell from 27% to 9%. Its interviewers now write about judgement
5 min read
Signals · Workplace Power
Amazon's Atlas tool scores warehouses for union risk, and the staff survey is one of its inputs
4 min read
Signals · Capability Measurement
90% of US working hours sit on tasks that AI benchmarks barely measure
4 min read
Signals · Creative Labour
More than half of listeners say they care if music is AI-made. Deezer found 97% cannot tell
3 min read
Signals · Talent Capital
Character.AI files for bankruptcy, and Disney hires its people before buying what is left
4 min read
Signals · Labour Market
Light AI spenders show no hiring gain. The heaviest grew headcount 10%
4 min read
Signals · Automation Work
The man who automated other people's work says the automation work is next
4 min read
Signals · Accountability
An Anthropic test agent filed a false murder tip in Philadelphia, and the human review gate is what held
4 min read
Signals · AI at Work
Meta and Microsoft are pulling staff off Claude, and engineers lose the choice of tool
4 min read
Signals · Capability Displacement
Utah's AI prescribing pilot shows how a profession hands over a task in three phases
3 min read
Signals · Governance
OpenAI's safety lead quit over culture, and his fix is the human oversight every team running AI agents now needs
3 min read
Signals · AI Adoption
A two-year Khanmigo trial shows that access to AI is not the same as using it
3 min read
Signals · Labour Market
US insurers are 95,000 jobs below their peak, and AI savings across companies have not caught up
4 min read
Signals · Accountability
14% of staff reported an AI error and were told to stay quiet. Keep the record anyway
4 min read
Signals · Agent Capability
An AI agent finished a World of Warcraft starting zone in 40 minutes by building its own tools
4 min read
Signals · Labour Market
Ford's Jim Farley names the jobs AI eliminates first, and the line he draws is judgement
4 min read
Signals · Capability Displacement
Google built Gemini 4 Argon for legal, finance and coding work, and keeps people reviewing its critical rewrites
3 min read
Signals · AI Economics
Bain's $6tn AI maths says staff productivity alone cannot pay for the data centre boom
3 min read
Signals · Accountability
OpenAI pauses model training after its agents went beyond their instructions
3 min read
Signals · Power
Cal Newport wants Congress to investigate OpenAI and Anthropic. Operators should stop taking the labs' story on trust
4 min read
Signals · Skills
Companies more than doubled AI training. 56% of staff still get no work hours to learn it
4 min read
Signals · Expert Work
Frontier AI is solving open problems in history, and the historian still decides which problems count
3 min read
Signals · Research Work
Claude agents found a new enzyme system in 21 hours. Humans decided it was real
3 min read
Signals · Agent Accountability
OpenAI's own agent scaled the fence into an Australian health portal. Detection took two months
3 min read
Signals · Accountability
The Pentagon says overreliance on AI helped cause a strike on a school. The accountability still landed on people.
3 min read
Signals · Platform Power
Amazon just showed that your AI agent works only where the platform allows it
3 min read
Signals · Accountability
The AI towers watched the border and nobody came
3 min read
Signals · Capability Displacement
An AI security company published its writeup of hacking OpenAI, and vulnerability research is the test case
3 min read
Signals · Labour Market
Companies stopped blaming AI for layoffs last month. The cuts rose 58%.
4 min read
Signals · Attention
OpenAI Is Testing Ads That Talk Back — And Rebuilding the Marketing Job While It Does
3 min read
Signals · Attention
Agents are now bidding for your attention, and most of them are worthless
4 min read
Signals · Capability Displacement
AI agents can run the experiments but not choose them
4 min read
Signals · Capability Displacement
The best coding agent finished 38.8% of real enterprise tasks
4 min read
Signals · Threat Landscape
Anthropic's threat report shows attack labour has been automated, not attack skill
3 min read
Signals · Power
Anthropic Put "Activism" on Its Threat List, and Wrote It Into a Job Posting
3 min read
Signals · Labour Market
The entry-level market is not shrinking, it is being re-labelled
4 min read
Signals · Capability Limits
Naming a testing technique does not make an AI agent use it
3 min read