AI + Careers Briefing — October 21, 2026

Today's stories share one idea: AI is moving from experiment to something people now have to measure. A new survey finds executives are twice as likely as their staff to trust AI over their own judgment. Engineering leaders are checking whether the money they spend on AI tools per developer is paying off. Anthropic is pitching lawmakers a bill that sets out when a court could stop a model's release. And OpenAI's purchase of Orchestra Bio takes AI agents into drug development. For job seekers, the common thread is proof: being able to show what your judgment and your AI use actually produced.

Do executives trust AI more than their own judgment?

Executives are twice as likely as rank-and-file workers to say they trust AI over their own judgment, according to a survey from Charter, a workplace research and media company, reported by Axios.

In the survey, 32.8% of executives said they trust AI more than their own judgment, versus 16.3% of individual contributors. Among managers, 34.8% said the same. Charter surveyed 5,002 U.S. office workers in August, all at companies with at least 100 employees and all saying they use AI at work. The sample included 2,952 individual contributors, 1,547 managers and 503 executives, a group that covers vice presidents, C-suite executives, owners and founders.

Report coauthor Brian Elliott, a former Slack executive who is now executive-in-residence at Charter, said executives face pressure to answer many questions quickly, and AI can help with that in the short term. Axios adds a caveat: the survey measures self-reported trust, not whether executives follow AI's advice on consequential decisions.

The report was broadly positive on AI, with 42.7% of AI users saying it improved both their work and personal lives. Separate interviews with 91 more workers surfaced a tension: the better workers get with AI, the more they realize how much can be done without them. Elliott said, "The odds that an AI CEO shows up and takes over a company are excruciatingly low."

Why it matters for job seekers: The survey does not say how trust in AI affects hiring or promotion, so treat what follows as our analysis. Reviewing the findings, we would expect the people evaluating you to be comfortable with AI-assisted work, and Elliott said AI is better at an executive summary than at the specific problem in front of you. That points to a practical move: show where your judgment adds something a summary cannot, such as a decision you made, the context behind it and the result. Our guide to job searching in the AI era covers how to present that.

Source: Axios

What is the return on software engineers' AI spending?

Companies are spending from roughly $150 to $3,000 per developer per month on AI tools, and the return is still unclear at many of them, according to The Pragmatic Engineer's look at 27 companies.

In an Oct. 20 newsletter, author Gergely Orosz reports that company size and AI spend are unrelated. He says spending peaked in summer 2026 and then roughly halved for companies on API plans. One quant trading shop with about 2,000 developers spent $8 million in September, up from $500,000 in January, and caps employee AI spend at $25,000. Results vary. A large publicly traded company with 5,000-plus developers saw weekly pull requests per developer rise from 4 to 8. A Series D hospitality company with about 200 engineers, capped at $1,500 per developer, reported pull request frequency unchanged. A publicly traded SaaS company with about 600 engineers saw pull request volume rise but merged velocity stay flat. Separately, CodeRabbit data cited in the piece shows developers using its AI code review tool increased weekly pull request output by 113% year over year. Parts of the article, including the Big Tech and ROI analysis sections, are behind a paywall and are not covered here.

Why it matters for job seekers: Employers are watching AI spend per engineer, and the numbers above show output measures like pull requests do not always translate into faster delivery. We would expect interviewers to ask how you use AI tools and what changed as a result. Reviewing this report, we would prepare examples that show shipped outcomes, not just more code, and see our guide to the AI skills salary premium for how employers price these skills.

Source: The Pragmatic Engineer

What is Anthropic's proposed AI safety bill?

It is a draft bill, not law, that Anthropic is pitching to lawmakers and that would let a court block the release of an AI model if it poses a foreseeable, imminent risk of more than 50 deaths or more than $1 billion in property damage, according to TechCrunch.

The reported risks must stem from the creation or release of a biological weapon, the model evading its developer's controls, or a cyberattack. Independent evaluators registered with the U.S. Commerce Department would assess models. Labs could pay them, as long as the payment does not depend on the results of the evaluations. An evaluator who identifies an imminent risk would have 72 hours to notify the Commerce Department.

The draft would forbid any federal agency from halting a model's distribution on its own. Instead, Commerce would have to ask a Washington, D.C. judge for a court order. An Anthropic spokesperson told Politico, as quoted by TechCrunch, that the draft reflects proposals in the company's Advanced AI Framework, which it has pushed with federal and state lawmakers since June.

Why it matters for job seekers: This is only a proposal, so nothing here is required yet. But the draft shows where the work could be: registered evaluators, risk assessment, and the documentation needed to show a model was tested. Reviewing the proposal, we would read it as a signal to build evidence of risk analysis and clear technical writing alongside your core skills. Our guide to AI-proof career skills covers how to show that kind of judgment. Watch whether the bill gains sponsors before planning around it.

Source: TechCrunch

Why did OpenAI buy Orchestra Bio?

OpenAI bought Orchestra Bio to expand its life sciences business, folding the startup's AI agents for drug-development operations into its Rosalind model, according to TechCrunch.

Orchestra Bio is a four-year-old AI biotech software startup. Deal terms were not disclosed. TechCrunch reported the company had reportedly raised approximately $20 million as of 2025 from investors including Andreessen Horowitz, Atria and 8VC. Its agents help biotech companies with drug-discovery operations, from financial tracking to scheduling clinical trials.

OpenAI plans to integrate Orchestra's team and technology into Rosalind and Rosalind Workbench, a tool that helps researchers plan experiments and interpret results. TechCrunch said OpenAI reports Moderna and Novo Nordisk are among Rosalind's customers.

The Decoder, citing OpenAI's announcement, reported that Orchestra's agents track trials, alert teams to timeline disruptions such as failed production batches, and recommend next steps. That would let scientists focus on research and strategic decisions rather than coordination. The Decoder also said developing a drug in the US takes 10 to 15 years on average.

Why it matters for job seekers: Reviewing these reports, we see AI labs moving from general tools into specific industries, with drug development the latest. The work Orchestra's agents are described as handling is coordination: tracking trials, flagging delays, suggesting next steps. For people in biology, chemistry, clinical operations or regulatory roles, that points to value in knowing the domain well enough to direct and check AI tools. Neither report gives hiring numbers, so treat this as a direction, not a forecast. If you are weighing a move into this kind of work, our roundup of online courses for a career change is a starting point.

Source: TechCrunch and The Decoder

What to watch

• Whether Anthropic's draft safety bill finds sponsors in Congress, and how other AI labs respond to it.
• Whether more companies cap AI tool spending per developer, as several in The Pragmatic Engineer's sample already do.
• How OpenAI folds Orchestra's agents into Rosalind, and which drugmakers sign up.