WorkThe story, in brief

Why AI companies are really pumping the brakes on their models

80% of workers say AI makes them more productive. Only 37% of companies see it in earnings—and that number hasn't moved in a year.

Illustration of a transparent lens revealing connected networks across layers of paper.
Exploring the next frontier of AI research.AI illustration by KeyNews
The KeyNews take

Why it matters

Opinion piece arguing that AI safety rhetoric masks two uncomfortable truths: the productivity-profitability gap isn't closing, and frontier labs are slowing development to protect margins rather than genuinely secure systems. Framed as industry commentary on AI's unfulfilled economic promises.

The key facts

12 to know
  1. McKinsey survey: 80% of people report AI productivity gains; only 37% of companies see earnings impact (unchanged year-over-year)

  2. Anthropic claims adjusted operating income profitability for second consecutive quarter—but not bottom-line profitability

  3. No frontier-model company (Anthropic, OpenAI) is profitable on GAAP basis; only NVIDIA among AI ecosystem profiting

  4. Article cites Jacob Coxon departure from Anthropic and Evan Hubinger's >10% existential risk claim as safety-narrative trigger

  5. References Hugging Face breach and German wiki agent attacks as backdrop to CEO safety calls

  6. Argues Chinese AI vendors and open-source developers won't observe slowdown despite US lab rhetoric

  7. McKinsey survey: 80% of people report AI improves productivity; only 37% of companies see earnings impact (unchanged for one year)

  8. Anthropic claims second consecutive quarter of positive adjusted operating income (AOI), but not bottom-line profitable

  9. OpenAI, Anthropic, and xAI leaders (Amodei, Altman, Musk) called for frontier model slowdown citing safety

  10. No frontier-model companies are profitable; NVIDIA is the primary commercial beneficiary

  11. Article cites Hugging Face hack, German wiki site attacks, and four known Anthropic hacking attempts as safety triggers

  12. Author argues safety narrative masks inability to deliver AI productivity fantasy and may be competitive/financial strategy

Go to the source

Computerworldcomputerworld.com

Publisher excerpt: There we were, listening to AI leaders doing their usual spiel: AI is great! AI will cure cancer! AI impact will be “unprecedented, perhaps 10x of the Industrial Revolution at 10x the speed”! AI will find a final answer to “Why do socks disappear in washing machines?” (Well, maybe not the last one.…
Read original report
Back to today's editionMore work news

The wider picture

View all
Illustration of two anonymous hands arranging task cards around an amber tool on a shared desk.
AI illustration by KeyNews
Work01

AI in finance must be policed differently

As agentic AI enters financial services, regulators face a choice: adapt existing oversight or risk strangling innovation. This opinion argues for AI-native regulation rather than forcing agents into legacy compliance.

Financial Times Technology
Illustration of two anonymous hands arranging task cards around an amber tool on a shared desk.
AI illustration by KeyNews
Work02

US and China agree to dialogue on AI ahead of Trump-Xi meeting

Policy and regulatory coordination on AI between the world's two largest AI powers is emerging as a formal diplomatic track. This affects how practitioners navigate export controls, chip sanctions, and cross-border AI deployment.

Financial Times Technology
Illustration of two anonymous hands arranging task cards around an amber tool on a shared desk.
AI illustration by KeyNews
Work03

Trump says he’s planning to create an ‘AI Force’ and hire a new AI czar

Government AI strategy and personnel moves directly shape how AI companies navigate regulation, funding, and deployment priorities. A new White House AI czar and dedicated military-style AI unit signals serious federal commitment to AI competitiveness and could reshape vendor relationships, export controls, and domestic AI investment.

SiliconAngle