News

AI + Careers Briefing — November 21, 2026

AI + Careers Briefing — November 21, 2026

Today's stories are about checking what sits underneath the surface: a proof, a tool and a new contact. An Anthropic mathematician claims to have used Claude to prove the Riemann hypothesis, a claim that BBC News and TechCrunch report mathematicians still need time to digest. A House report reviewed by CNBC says the AI coding platform Cursor is considering a move away from Chinese-developed AI models. And security experts told TechCrunch that many more people in AI than just founders and CEOs are at risk of honey-trap schemes.

Did AI just prove the Riemann hypothesis?

Not confirmed. An Anthropic employee and mathematician, Levent Alpöge, claims to have used the company's Claude chatbot to prove the famed Riemann hypothesis, according to BBC News. BBC reports that Alpöge said he spent about six weeks using Claude to create a detailed proof, and that the proof would be among the most significant in history if found to be correct. TechCrunch says he appears to have solved one of the most important problems in mathematics, and notes the hypothesis dates back to 1859 and is among the Millennium Prize math problems.

What has been checked so far

The two outlets stress different things. TechCrunch reports that Alpöge's result is a Lean-compilable solution. It describes Lean as an open-source coding language used to express mathematical ideas as code, allowing proofs to be formally verified. Several participants in a mathematicians' chat, including MIT mathematician Andrew Sutherland, say they have confirmed the solution compiles, TechCrunch says. It adds that it will be a while before the mathematics community has a handle on the proof, and that humans still need to understand the result and explain it to each other.

BBC reports that the complete textual proof in Lean is over 250,000 pages long, and that it was found using a non-public version of Claude. BBC quotes the mathematician Terence Tao, who said on Friday that verification is still some way off:

"I imagine the digestion process will take some time, though in this case at least we have a human author to communicate with to assist with this task," Tao said.

A second claim, with no AI

TechCrunch also reports that Ravi Vakil, a Stanford professor and the president of the American Mathematics Association, posted a proof of the General Riemann hypothesis written by Stanford postdoc Alessandro Masullo. Vakil noted that it is all human, with no AI. TechCrunch says that without being expressed in Lean, it is too early to say whether that proof checks out. Masullo said his proof is the result of 12 years of work.

BBC adds context on earlier AI claims. It says OpenAI claimed in September that employees using a non-public version of its tools solved part of a problem in the Navier-Stokes equations, and that this proof has yet to be confirmed as accurate.

What this means for job seekers: This section is our analysis, based only on the reporting above. Nothing in either article discusses hiring or jobs, so these are practical takeaways, not predictions.

Both articles show the same pattern. Alpöge worked with Claude for about six weeks, yet the story does not end at the AI's output. The claim still depends on people checking it, and BBC quotes Tao pointing to the value of having a human author to communicate with. If you work in research or other knowledge work, being able to explain and check what an AI tool produced appears to matter as much as producing it. Our guide to AI-proof career skills is a place to start.

Three practical steps follow from the reporting:

  • Describe your AI work in interviews as a process: what you asked the tool, what you checked, and what you could explain afterward.
  • Keep your own record of how a result was reached. TechCrunch notes the hard part now is humans understanding and explaining the result.
  • Treat claims, including your own, as unconfirmed until independently checked. Both outlets hedge the proof with words like "claims" and "may have."

Source: TechCrunch, BBC News.

Is Cursor moving away from Chinese AI models?

Cursor is considering it. The AI coding platform is weighing a move away from Chinese AI models because of "geopolitical risk," according to a House report reviewed by CNBC. The report, produced by the House Committee on Homeland Security and the House Select Committee on China, says Cursor plans to explore phasing out open-weight models developed in the People's Republic of China.

CNBC reports that Cursor, which SpaceX acquired at an implied $60 billion equity value in August, had built its coding model Composer 2 on top of Chinese lab Moonshot AI's Kimi K2.5. The report says Cursor informed the committees of its plans. Airbnb and DoorDash also engaged with the investigation about their use of Chinese AI models.

Cursor, SpaceX and DoorDash did not respond to a CNBC request for comment, and Airbnb declined to comment.

Why Chinese models caught on

The House report follows a months-long investigation into the adoption of Chinese AI models in the American tech stack. It found that cost, customization and near-frontier performance were driving adoption, and that the U.S. lacks a competitive open-weight alternative at "price and performance the market demands."

The report also warns about the long game. "Beijing is now seeking to turn those technological gains into global market share and long-term dependence," the report says.

Among its recommendations are a "national American open-weight initiative" and building on an order, made public in October, that banned DeepSeek from federal civilian agency systems. The report says that order should be extended to prohibit "PRC-developed AI models" in federal information systems and in performance of federal contracts.

What this means for job seekers: This section is our analysis, based only on the reporting above. The CNBC report is about one company's plans and a congressional report, and it does not describe any effect on hiring or on individual developers.

Still, one detail is worth noting. Cursor's own coding model was built on top of another lab's model, and Cursor is now considering changing that foundation. If the model underneath a tool can change for geopolitical reasons, a skill that depends on one tool's exact behavior is the riskier thing to feature on a resume.

Practical steps follow from that. Describe your AI work by the task and the result rather than only the product name, and show you can work across more than one assistant. In interviews, ask which AI tools and models the team is approved to use, and whether those rules could change. The report recommends barring PRC-developed models from federal contracts, so anyone targeting that kind of work should ask the same question early.

For more on how AI fluency shows up in pay, see our AI skills salary premium guide.

Source: CNBC.

Are people working in AI being warned about honey-trap schemes?

Yes. Security experts told TechCrunch that many more people in AI than just founders and CEOs are at risk of honey-trap schemes. TechCrunch framed the story around the fall-out for Nvidia over Jensen Huang's alleged honey-trap scandal, which it said is still unfolding.

"Nobody briefs founders on counterintelligence," said Aaron Brown, a former senior CIA officer and current founder of the software company Lumbra. TechCrunch reports that honey traps have historically been associated with politicians, the military and national intelligence, but that many of the most valuable secrets now sit inside companies building chips, models and infrastructure.

The risk is not limited to executives. The WSJ reported that OpenAI started warning its employees about outsiders trying to cozy up to them in search of company secrets, according to TechCrunch. Anyone with access to sensitive information, from board members to executive assistants, can be a target.

How the approaches work

Typically, the attackers are not lovers, TechCrunch says. Brown said friendship is the oldest honey-trap scheme around. Honey traps gain access to their targets as investors, customers, recruiters and consultancies, and their requests look legitimate on paper. "They need you to feel comfortable talking," Brown said.

TechCrunch also reports that AI has made honey-trapping easier and cheaper. It says planning a scheme used to take months or years, while now it is easy to use AI to research a target and generate fake social profiles, fake backgrounds and even fake voice calls. "AI can help sustain an entire fictional backstory," said Perry Carpenter, a criminologist and chief deception strategist at KnowBe4.

Warning signs the experts describe

Carpenter said a good first step is looking for unusual patterns, such as coincidental encounters that happen a little too often. TechCrunch lists other signs: persistent curiosity about work and colleagues, requests for secrecy about meeting up, and pressing for introductions or information. Bigger red flags include requests that keep growing and push past what you would usually be comfortable doing.

Brown offered a gut check: "Would this person still be around if you lost your job tomorrow?" Carpenter suggests a clear boundary between personal and professional life. Brown also said, "Paranoia doesn’t help. Recognizing you’re a target does."

What this means for job seekers: This section is our analysis, based only on the reporting above. The article does not say employers are changing hiring or screening because of this warning.

First, treat security awareness as part of working in AI. Brown's point that nobody briefs founders suggests training may be uneven, and the OpenAI warning shows at least one employer is addressing outsiders. In interviews, it is reasonable to ask whether new hires receive any guidance on approaches from outsiders.

Second, recruiters appear on the list of roles honey traps use. That does not make recruiters suspect. It does mean that persistent questions about your current work or colleagues, beyond what a role requires, match the red flags the experts describe. Share what is needed to assess fit, and keep confidential details out.

Third, ask about culture. The experts tie safety to whether employees feel comfortable raising concerns and challenging unusual requests. Asking how concerns get escalated tells you something about a team before you accept an offer.

For broader preparation in AI-affected fields, see our AI-proof career skills guide.

Source: TechCrunch.

What to watch

  • How mathematicians' review of the claimed Riemann proof goes. BBC News quoted Tao saying the digestion process will take some time.
  • Whether Cursor follows through. The House report says it plans to explore phasing out open-weight models developed in the People's Republic of China, and Cursor did not respond to CNBC's request for comment.
  • Whether more AI employers warn staff about outside approaches, as the WSJ reported OpenAI has, according to TechCrunch.
Posted in
News

About the author

Julian G. — Writer & Editor

Julian G. is a web developer who has run job4travelers.com and udreamjob.com since 2019. He writes about remote work, job searching, career strategy, and travel — topics he's followed for years as both a practitioner and a reader. Some posts draw on personal experience; others synthesize research from primary sources. Every post is reviewed and edited by him before publishing.

Related Posts

Job Opportunities

Browse all opportunities →