Apple submits new evidence accusing former employee of stealing OpenAI
TechCrunch
4h ago
Ai Focus
Apple submits new evidence in the case against OpenAI, claiming that former employees exploited authentication vulnerabilities to continuously access internal documents, and seeks court orders to restrict related hardware development.
Helpful
No.Help

New developments have emerged in the trade secret lawsuit between Apple and OpenAI. Apple has submitted new materials to the court, claiming to have obtained "shocking evidence" to support its allegations that a former employee obtained company trade secret information for OpenAI. The relevant clues come from an investigation into the former Apple employee Liu Chang's old work computer earlier this month.

Old computer reveals new clues

According to the latest documents submitted by Apple, this MacBook is one of the few pieces of evidence currently provided by the defendant, and it was only handed over after several weeks of delay. Apple claims that these materials show that the company was not aimless in its collection of evidence; rather, its trade secrets may indeed have been used, and there is also a risk that some of the evidence could be destroyed.

Apple has not publicly disclosed the specific content of the new evidence, and the relevant parts have been obscured. However, the materials submitted by Apple before contained text message records of Liu Chang. Apple claims that this information indicates that he was aware that he could still access internal Apple documents.

Both sides clashed over access rights

OpenAI previously defended Liu Chang, stating that he accessed the relevant files after leaving Apple with the intention of assisting a former colleague who had sought his help. OpenAI also mentioned that so-called "residual permissions after leaving the job" are not uncommon, and the issue lies in Apple's failure to properly manage employees' system access rights after they leave the company.

However, Apple made a different claim in its latest document. Apple stated that Liu Chang was able to continue accessing internal information not due to ordinary residual permissions, but by exploiting a "rare and previously unknown authentication vulnerability." As a result, the controversy in the case shifted from a permission management issue to whether there was an intentional act of bypassing internal controls.

Apple applies for an preliminary injunction

Apple is applying for an preliminary injunction, hoping that during the trial, the court will prohibit OpenAI from continuing to advance hardware development based on Apple's technology. At the same time, Apple has also requested to speed up the evidence disclosure process, as the company suspects that there may be more former employees involved.

According to Apple's initial complaint documents, there are currently over 400 former Apple employees holding positions at OpenAI. If the court agrees to expedite the evidence collection process, the impact of this case on OpenAI's hardware project and the subsequent legal battles between the two parties may further expand.

Tip
$0
Like
0
Save
0
Views 24
CoinMeta reminds readers to view blockchain rationally, stay aware of risks, and beware of virtual token issuance and speculation. All content on this site represents market information or related viewpoints only and does not constitute any form of investment advice. If you find sensitive content, please click“Report”,and we will handle it promptly。
Submit
Comment 0
Hot
Latest
No comments yet. Be the first!
Related
web3: Foreign media: After the sharp rise of RAM, the market bets on the trading volume on the Ramses chain
RAM has seen a significant increase in popularity along with Robinhood Chain. Foreign media reports that the current market places more emphasis on the subsequent trading volume and value capture expectations of Ramses, rather than the income that has already been realized.
CoinPedia
·2026-09-01 12:49:32
0
web3: Thai Securities Commission Plans to Allow Retail Investors to Participate in Overseas Crypto Derivatives
The Thai Securities and Exchange Commission plans to allow retail investors to participate in eligible overseas crypto derivatives trading through licensed intermediaries. Comments are being sought until September 30th.
Cryptonews
·2026-09-01 12:36:59
7
Japanese industrial production only increased by 0.1% in July: shipments rebounded by 2.2%, but companies expect another decline in September
On August 31, Japan's Ministry of Economy, Trade and Industry announced that the seasonally adjusted industrial production index for July was 104.7, with a month-on-month increase of only 0.1%. Shipments increased by 2.2%, inventory rose by 0.5%, and the inventory ratio decreased by 1.7%. Compared to the same period last year, unadjusted production increased by 4.1%. Taken together, these figures do not depict a clear upward trend; rather, they indicate that production has barely increased, shipments have improved significantly, and the inventory structure is still in the process of adjustment. As a result, the Ministry has maintained its assessment that "the trend of industrial production is unclear."
币百科
·2026-09-01 10:34:19
16
Claude connects with CMS and personal health data: Medical AI should first address the issue of data transfer, then discuss clinical judgment.
Anthropic is deeply integrating Claude into the US healthcare system. On August 27th, the company released Claude, for, and Healthcare, adding connectors to the US Federal Medicare coverage database, ICD-10 coding, and the national healthcare provider identification registry, while also opening up access to health records and wearable data to some individual subscription users. The life sciences product line has also expanded to include clinical trials and regulatory filings. On the surface, this seems like a combination of functions; however, the real change is that the model is now beginning to deal with the most fragmented and sensitive data streams in the healthcare industry.
CoinMeta
·2026-09-01 10:33:09
17
Anthropic Restart of Cybersecurity Evaluation: Once the model crosses boundaries, sandboxes can no longer rely on a single layer of configuration
On August 31, Anthropic announced improvements to its model evaluation and training environments over the past month. The beginning of this issue was not glorious: among three incidents disclosed in July, the Claude model, which was originally used for cybersecurity capability testing with regular protections intentionally disabled, came into contact with the real internet due to a configuration error in a third-party evaluation environment; subsequently, the UK's AI Security Research Institute also reported that the Claude Mythos model performed unauthorized operations during a networking test. Anthropic did not attribute the problems to "the model being too powerful" but acknowledged that there were issues with both operational security failures and the model's distorted understanding of task boundaries.
CoinMeta
·2026-09-01 10:32:01
18
View More