Tech
tech
Jon Keegan

Musk’s xAI sues former employee, alleging theft of trade secrets

Elon Musk’s xAI has filed suit in Northern California federal court against a former engineer, alleging that he stole trade secrets from the company just before leaving for a job at OpenAI.

Xuechen Li started working for xAI in February 2024 after receiving his Ph.D. in computer science from Stanford University.

In late July, Li sold $7 million worth of his xAI shares to the company as part of a liquidity grant for employees.

The complaint alleges that once the sale was complete, Li illegally copied xAI files:

“On July 25, 2025 — the same day he concluded his second sale of equity and had millions in cash on hand — Defendant betrayed the trust and faith xAI had placed in him by willfully and maliciously copying xAI Confidential Information (as defined in the Agreement) and trade secrets from his xAI-issued laptop to one or more non-xAI physical or online storage systems within his personal control (collectively, ‘Personal System’).”

Three days later, Li resigned to take a job at OpenAI.

Reuters wrote that Li didn’t immediately respond to its request for comment.

The complaint also alleges:

“Defendant, with his attorney present, admitted in a handwritten document he provided to xAI that he misappropriated xAI’s Confidential Information and trade secrets, and again, with his attorney present, admitted verbally during in-person meetings with xAI that he engaged in such misappropriation and further admitted that he tried to hide his theft.”

The complaint also alleges Li withheld accounts and credentials from forensic investigators after handing over personal devices to the company.

The complaint alleges that once the sale was complete, Li illegally copied xAI files:

“On July 25, 2025 — the same day he concluded his second sale of equity and had millions in cash on hand — Defendant betrayed the trust and faith xAI had placed in him by willfully and maliciously copying xAI Confidential Information (as defined in the Agreement) and trade secrets from his xAI-issued laptop to one or more non-xAI physical or online storage systems within his personal control (collectively, ‘Personal System’).”

Three days later, Li resigned to take a job at OpenAI.

Reuters wrote that Li didn’t immediately respond to its request for comment.

The complaint also alleges:

“Defendant, with his attorney present, admitted in a handwritten document he provided to xAI that he misappropriated xAI’s Confidential Information and trade secrets, and again, with his attorney present, admitted verbally during in-person meetings with xAI that he engaged in such misappropriation and further admitted that he tried to hide his theft.”

The complaint also alleges Li withheld accounts and credentials from forensic investigators after handing over personal devices to the company.

More Tech

See all Tech
tech

Report: Google DeepMind builds “strike team” to catch up to Anthropic models

Anthropic’s recent momentum, powered by the success of its popular Claude Code tool, is turning up the heat among its AI competitors — not only for its AI startup peer OpenAI, but also with established Big Tech giants like Google.

The Information reports that within Google DeepMind, a “strike team” has been assembled to make a serious push to improve Gemini’s coding capabilities. According to the report, leaders within Google, including cofounder Sergey Brin, are sounding the alarm after determining that Anthropic’s Claude has superior coding skills. The new team’s goal is to create a AI system that can improve itself.

“To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers,” Brin wrote in a recent memo to DeepMind staff.

The Information reports that within Google DeepMind, a “strike team” has been assembled to make a serious push to improve Gemini’s coding capabilities. According to the report, leaders within Google, including cofounder Sergey Brin, are sounding the alarm after determining that Anthropic’s Claude has superior coding skills. The new team’s goal is to create a AI system that can improve itself.

“To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers,” Brin wrote in a recent memo to DeepMind staff.

$0

Tesla’s federal tax bill last year was once again $0, Reuters reports. While past losses and green energy credits helped shrink the bill, Reuters found that Tesla also leaned on a classic corporate maneuver: offshore profit-shifting. By routing intellectual property rights through paper-only subsidiaries in the Netherlands and Singapore, Tesla effectively parked $18 billion in profits overseas between 2023 and early 2025. The entirely legal setup saved Tesla an estimated $400 million in US taxes. Not bad for a company whose CEO is not a fan of “shady” tax loopholes.

tech

Report: NSA is currently using Anthropic’s unreleased Mythos model

According to the Pentagon, Anthropic’s AI tools are a national security supply chain risk, and have been banned for defense applications.

But a new report says the National Security Agency, which operates as a part of the Pentagon, is currently busy using Anthropic’s new, unreleased AI model, Mythos.

Axios reports that Mythos’ reputed advanced offensive cyber capabilities have compelled the NSA to begin using it, despite the public blacklisting from the Pentagon, which Anthropic is suing the US government over.

Anthropic has granted access to a small number of trusted partners to test and prepare for the expected explosion of vulnerabilities to be discovered using the new AI model. UK intelligence agencies have also reportedly gained access to Mythos.

Anthropic CEO Dario Amodei reportedly visited the White House last week to try and resolve the dispute on allowing wider use of the company’s technology in the federal government.

Axios reports that Mythos’ reputed advanced offensive cyber capabilities have compelled the NSA to begin using it, despite the public blacklisting from the Pentagon, which Anthropic is suing the US government over.

Anthropic has granted access to a small number of trusted partners to test and prepare for the expected explosion of vulnerabilities to be discovered using the new AI model. UK intelligence agencies have also reportedly gained access to Mythos.

Anthropic CEO Dario Amodei reportedly visited the White House last week to try and resolve the dispute on allowing wider use of the company’s technology in the federal government.

Latest Stories

Sherwood Media, LLC produces fresh and unique perspectives on topical financial news and is a fully owned subsidiary of Robinhood Markets, Inc., and any views expressed here do not necessarily reflect the views of any other Robinhood affiliate, including Robinhood Markets, Inc., Robinhood Financial LLC, Robinhood Securities, LLC, Robinhood Crypto, LLC, Robinhood Derivatives, LLC, or Robinhood Money, LLC. Futures and event contracts are offered through Robinhood Derivatives, LLC.