The Morning Download: Space Race Meets AI Race

Good morning. The original space race drove technological innovation through intense competition between superpowers. Today the race that matters stars a different set of superpowers...and it involves code.

SpaceX on Tuesday announced that it secured a $60 billion option to acquire AI coding tool Cursor, marking the latest effort by Elon Musk's rocket company to reinvent itself ahead of its planned offering, the WSJ's Micah Maidenberg and Angel Au-Yeung report. SpaceX acquired xAI earlier this year, folding it into the aerospace operation.

SpaceX said it is already working closely with Cursor on coding and AI, but a deal would further commit it to a race more unforgiving than the vacuum of space. Cursor makes a tool that allows developers to toggle between different AI models from different labs. Competition has increased as those same AI labs launched their own coding tools, most notably Anthropic’s Claude Code and OpenAI’s Codex. Last fall, Cursor launched its own AI model, called Composer.

SpaceX sees a potential Cursor acquisition as part of a larger strategy. The company is also planning to use its space assets to compete with AI heavyweights, asking regulators to deploy up to one million AI satellites to take on computing tasks, the Journal reports.

For those stuck on earth, the race for coding dominance is just one front in the broader battle for AI supremacy—a landscape that's shifting faster than ever. More on the latest AI model moves below.

Unauthorized users access Anthropic's powerful AI model. Anthropic has kept a tight grip on Claude Mythos, allowing access to a select group of organizations. But that has not stopped a small group of unauthorized users from accessing the model. Bloomberg reports that the group exploited a contractor's access to Anthropic and used internet research tools to locate the model. Anthropic says there's no evidence the breach extends beyond a vendor's environment.

Trump talks Anthropic. After ordering federal agencies to “IMMEDIATELY CEASE all use of Anthropic’s technology” in February, President Trump appears to have had a change of heart, telling CNBC that “it’s possible” that Anthropic will be back in the Defense Department. 

The administration called Anthropic a supply-chain risk after the company would not agree on how AI should be used by the Defense Department. But that was before Anthropic unveiled Claude Mythos, a powerful AI model that has equally spooked and intrigued Washington for its ability to identify and exploit software bugs.

Meta wants to track employees for AI. Meta Platforms is deploying tracking software on U.S. employee computers to capture mouse movements, clicks, and keystrokes for AI model training, Reuters reports. The tool will monitor work apps and take periodic screenshots to help build autonomous AI agents for workplace tasks.

Florida is launching a criminal investigation into OpenAI. The state's attorney general said the investigation focuses on messages between ChatGPT and the suspect of a mass shooting that killed two people at Florida State University last year, the WSJ reports.

When AI runs the shop. Startup Andon Labs rented commercial space in San Francisco, filled a bank account and then fired up an AI agent called Luna with instructions to make a profit. The result is Andon Market, the first boutique store run by AI. Of course there are quirks. For one there is a serious staffing problem (Luna screwed up the work schedule) and then there is the merchandise ordered by AI: "candles — so many candles," says the New York Times. 

Signs that entry-level hiring is picking up. A widely watched survey by the National Association of Colleges and Employers out Monday shows employers expect to boost new-graduate hires by 5.6% this spring from a year ago—a turnaround from their much grimmer forecasts last fall. In some cases, artificial intelligence is spurring hires by enabling companies to expand services and product lines, said NACE President Shawn VanDerziel.

Dutch Intel chief warns of China’s cyberspying. China’s cyber-espionage capabilities are now as sophisticated as the U.S.’s and are increasingly targeting Western defense industries, said the head of Dutch military intelligence. An accompanying report noted that Russia continues to pose the biggest and most direct threat to Europe and growing ties between Russia and China are fueling Moscow’s ambitions.

Anthropic's new policy opens door for Chinese rivals. Despite being denied commercial access to Anthropic’s tools, Chinese AI developers, through some easy hacks, have proven to be big fans of Claude. But that is changing with a new requirement that some customers submit government ID and a photo for verification. The Information reports that their temporary setback may be a gain for local alternatives like DeepSeek's V4 and Zhipu's coding tools.

A gold-hued skyscraper rising above Phnom Penh is a monument to the spoils generated by Cambodian cybergangs that have stolen billions of dollars from unsuspecting Americans and others, the WSJ reports. The skyscraper is being built by a company under sanctions by the U.S. Treasury Department for its alleged connection to scam operations that have cropped up across Cambodia.

Under escalating international pressure, Cambodia’s government is cracking down, exposing the magnitude of the industry. At the same time, the crackdown has prompted many large operations, some the size of small towns, to scatter, making them harder to track.

Three ships came under attack in the Strait of Hormuz as tensions flared in the waterway, after President Trump said the U.S. would extend its cease-fire with Iran and continue its blockade until Tehran presents “a unified proposal.” (WSJ)

As negotiations with Iran broke down Tuesday before they had even begun, President Trump asked aides whether the U.S. should resume attacks on the country, according to U.S. officials. (WSJ)

Virginia voters narrowly approved a measure to redraw the state’s congressional map ahead of the midterms, the Associated Press projected, giving a big boost to Democrats’ efforts to reclaim control of the U.S. House. (WSJ)

New data from two drugs showed it might be possible to keep pancreatic cancer in check for longer than ever before. One drug, developed by Revolution Medicines, shrank tumors in roughly half of people who used it as a first treatment. And an mRNA vaccine made by Germany-based BioNTech and Genentech kept most patients who responded to it alive six years. (WSJ)

The WSJ Tech Council brings together CIOs, CTOs and CISOs advancing innovation and shaping the future. Join this trusted community where tech executives connect with peers to explore emerging trends and gain the perspective they need to stay ahead of disruption.

Request Information

The WSJ CIO Journal Team is Steven Rosenbush, Isabelle Bousquette and Belle Lin.

The editor, Tom Loftus, can be reached at thomas.loftus@wsj.com.