A former Apple engineer allegedly accessed a confidential power-converter design after joining OpenAI and used it in AI work. Apple now says the case has taken a more serious turn: evidence may have been deliberately destroyed.
The escalating legal battle between Apple and OpenAI has entered a more serious phase, with Apple accusing the ChatGPT maker of failing to preserve—and potentially destroying—evidence connected to a former Apple engineer who allegedly accessed confidential Apple technical information after joining OpenAI.
The latest allegations were contained in a new filing in a U.S. federal court and reported by Bloomberg and Reuters. Apple says newly examined evidence from an OpenAI-issued MacBook provides additional support for its trade-secret claims.
This is no longer simply a dispute about employee mobility or whether former Apple engineers carried knowledge into their new jobs.
Apple is now alleging potential evidence destruction.
That could significantly raise the stakes of the case.
The Engineer at the Centre of the Dispute
The case centres on Chang Liu, a former Apple senior system electrical engineer who joined OpenAI in January 2026.
Apple alleges that after joining OpenAI, Liu accessed a confidential Apple power-converter circuit schematic containing technical information about Apple's hardware designs. According to Apple's latest filing, the document was downloaded from Apple's third-party cloud storage on March 7, 2026.
Apple says the schematic was not merely viewed accidentally.
Its latest filing alleges that Liu used the confidential information in his work at OpenAI and even used it in connection with an AI agent.
That allegation is particularly significant because it goes directly to Apple's central concern: whether proprietary Apple engineering information crossed into OpenAI's development environment.
The AI Angle Makes This Case Different
The most striking part of the allegations is not simply that a former Apple employee allegedly accessed a confidential document.
It is how Apple says the information was used.
According to the company's filing, Liu worked with an AI agent capable of interacting with engineering software and analysing circuit simulations.
Apple alleges that its confidential circuit information was used in this work.
That creates a new category of corporate trade-secret anxiety.
Historically, an employee might take a document, copy a design or explain a process to a colleague.
AI introduces another possibility:
What happens when confidential engineering knowledge is fed into an AI-assisted workflow?
Once proprietary information enters an AI-enabled development process, determining exactly where that information went—and what systems subsequently interacted with it—can become considerably more complicated.
The case therefore touches on an emerging problem for virtually every major technology company:
How do you prevent employees from taking confidential information into AI systems?
Apple Says the Evidence Trail Gets Worse
Apple's newest allegations go beyond the alleged access to its circuit design.
The company claims that evidence found during forensic examination of an OpenAI-provided MacBook suggests that Liu continued accessing Apple's cloud storage after leaving the company.
Apple says Liu's access continued into April 2026 and that files were downloaded during that period.
This is important because OpenAI has previously argued that some of the data Apple says was accessed may have been available because of Apple's own account and device-security practices.
OpenAI's legal strategy has included arguing that Apple's security and employee offboarding procedures weakened its trade-secret claims. In particular, OpenAI has argued that Apple failed to take reasonable measures to protect information it now characterises as trade secrets.
So the dispute is becoming a battle over both sides' security practices.
Then Comes the Evidence-Destruction Allegation
This is potentially the most explosive part.
Apple alleges that after Liu became aware of Apple's investigation, he exchanged messages with an OpenAI colleague concerning Apple-issued devices.
According to Apple's filing, Liu instructed the colleague to have devices “restored” and subsequently used again.
Apple characterises this as potentially anti-forensic behaviour, because restoring a device can overwrite or eliminate forensic traces that investigators might otherwise recover.
The company says this constitutes evidence destruction.
That is an allegation—not an established finding by the court.
But if Apple can prove that evidence was intentionally destroyed after the parties became aware of the litigation or investigation, the consequences could be considerably more serious than the underlying trade-secret dispute.
Why Evidence Preservation Matters So Much
Once litigation is reasonably anticipated, companies and individuals involved in a dispute generally have obligations to preserve potentially relevant evidence.
That can include:
- emails;
- text messages;
- laptops;
- cloud-storage records;
- source files;
- engineering documents;
- internal communications;
- device logs;
- collaboration platforms;
- AI-related activity.
The purpose is straightforward.
A court needs to determine what actually happened.
If potentially relevant evidence disappears, the opposing party can argue that the missing information might have been damaging.
That is why Apple's accusation of evidence destruction could become a major issue in its request for expedited discovery.
Apple Wants the Court to Move Faster
Apple is asking the court for expedited fact-finding and broader discovery.
The company argues that the new evidence demonstrates why it needs access to additional devices, accounts and records before the case proceeds further.
The logic is straightforward:
If one forensic examination uncovered evidence that Apple says contradicts OpenAI's account, Apple wants to know what else might be found.
That could potentially expand the investigation beyond Liu.
Apple has already sent preservation notices to roughly 40 former employees who left Apple for OpenAI, according to reporting on the broader dispute.
That illustrates how wide the conflict could become.
OpenAI Has Strongly Rejected Apple's Case
OpenAI has not accepted Apple's allegations.
Earlier in August, OpenAI publicly described Apple's lawsuit as baseless and argued that Apple was mischaracterising what happened with former employees.
OpenAI has maintained that it does not want Apple's trade secrets and has challenged Apple's description of the information at issue.
OpenAI has also asked the court to dismiss Apple's lawsuit.
Its legal strategy has focused partly on Apple's own information-security practices, arguing that weaknesses in Apple's offboarding and account-management systems undermine Apple's assertion that the information qualifies as legally protected trade secrets.
So far, therefore, there are two sharply competing narratives.
Apple's version:
Former employees accessed confidential information, used it at OpenAI and potentially destroyed evidence that could prove what happened.
OpenAI's version:
Apple's allegations are overstated, its security practices contributed to the access, and the information does not establish the trade-secret case Apple claims.
The court ultimately has to determine which version is supported by evidence.
This Is Bigger Than One Employee
The lawsuit comes at a strategically sensitive moment.
OpenAI has been moving toward hardware, including ambitions that could eventually put its products into direct competition with Apple's ecosystem.
That makes Apple's trade-secret concerns particularly significant.
The dispute is not simply about an employee moving from one technology company to another.
It concerns the movement of hardware engineering expertise between two companies that could increasingly compete over the future of consumer computing.
Apple's concern is therefore understandable from a competitive standpoint.
If confidential engineering information connected to future hardware products reaches a potential competitor, even indirectly, the commercial consequences could be enormous.
The AI Industry Has a New Trade-Secret Problem
This case also highlights something much larger than Apple and OpenAI.
AI is changing how engineers work.
An engineer can now use an AI agent to:
- analyse circuit designs;
- run simulations;
- optimise parameters;
- write engineering code;
- troubleshoot technical problems;
- interpret test results;
- automate repetitive engineering tasks.
That creates extraordinary productivity opportunities.
It also creates extraordinary information-security risks.
An employee does not necessarily need to email a 500-page confidential document to a competitor.
They could potentially expose sensitive information through an AI-assisted workflow.
And once an AI agent has interacted with proprietary information, companies may have to determine:
What did the system see?
Where was the information stored?
What logs exist?
Who had access?
Was the information transmitted externally?
Was it incorporated into another system?
Can the activity be reconstructed?
These questions are rapidly becoming central to corporate cybersecurity.
The Case Could Become a Warning to Every Technology Company
Apple and OpenAI are hardly alone in dealing with employee movement.
Engineers regularly move between:
- Apple;
- Google;
- Microsoft;
- Meta;
- OpenAI;
- Anthropic;
- Nvidia;
- startups;
- semiconductor companies.
They take their experience with them.
They cannot realistically erase everything they have learned during their careers.
That is why trade-secret law distinguishes between general skills and knowledge and genuinely confidential proprietary information.
A former engineer can use knowledge gained through years of professional experience.
The legal problem begins when confidential documents, source code, designs, customer information or other protected material are improperly taken or used.
The Apple–OpenAI case could therefore become an important test of where that boundary lies in an AI-driven workplace.
The Most Dangerous Part for OpenAI May Not Be the Trade-Secret Claim
If Apple's allegations are ultimately proven, the evidence-preservation issue could become particularly damaging.
A trade-secret dispute asks:
Was confidential information improperly obtained or used?
An evidence-preservation dispute asks something potentially more uncomfortable:
Did someone subsequently attempt to prevent investigators or the court from discovering what happened?
Those are very different questions.
Even if OpenAI ultimately succeeds in defending itself against Apple's underlying trade-secret claims, allegations concerning evidence preservation could create a separate legal problem for individuals involved.
At this stage, however, it is essential to emphasise:
Apple's allegations have not been established as facts by a court.
Apple and OpenAI Are Now Fighting Over the Future of Hardware
There is a fascinating strategic dimension beneath the litigation.
Apple dominates premium consumer hardware.
OpenAI dominates much of the public conversation around generative AI.
Now OpenAI wants to move deeper into devices and hardware.
That potentially puts the companies on a collision course.
Apple's strength is hardware, industrial design, silicon and tightly integrated software.
OpenAI's strength is AI models, agents and increasingly sophisticated AI interfaces.
The company that successfully combines those capabilities could shape the next generation of personal computing.
That makes engineering talent extraordinarily valuable.
And it makes proprietary engineering knowledge extraordinarily sensitive.
A New Era of Corporate Espionage Anxiety
The traditional image of industrial espionage involved someone walking out of a building with a hard drive or photographing a confidential document.
The AI era may look very different.
A developer could inadvertently expose proprietary information to an external AI system.
An engineer could use confidential specifications as context for an AI agent.
An employee could ask an AI assistant to optimise a proprietary design.
A departing employee could transfer years of institutional knowledge into an AI-powered workflow.
And companies may struggle to determine exactly what happened afterward.
That makes employee offboarding more important than ever.
When a senior engineer leaves, companies may need to consider not only:
“Did we recover the laptop?”
but also:
“What AI systems, cloud accounts, agents and external tools did this person have access to?”
The Apple–OpenAI Fight Could Set an Important Precedent
Whatever happens in court, this case is likely to be watched closely throughout Silicon Valley.
The technology industry is entering a period where:
talent moves faster, AI systems absorb more information, and corporate secrets are increasingly digital.
That combination creates a new category of legal risk.
Companies will have to rethink:
- employee access controls;
- offboarding procedures;
- cloud permissions;
- AI usage policies;
- device forensics;
- data-loss prevention;
- evidence preservation;
- confidential-information monitoring.
The traditional corporate security model was built around files, servers and email.
The AI workplace is considerably more complicated.
The Bigger Question: Who Owns the Knowledge?
At the heart of the dispute lies a question that will become increasingly important in the AI era:
Where does an employee's knowledge end and a company's intellectual property begin?
An engineer's expertise belongs to the engineer.
A company's confidential circuit design belongs to the company.
The challenge is determining the boundary between the two.
That boundary becomes even more complicated when an engineer works alongside an AI agent capable of turning instructions and technical information into new outputs.
The courts may increasingly be asked to determine exactly where that line lies.
Conclusion: The AI Talent War Has Become a Data War
Apple's latest filing dramatically raises the stakes in its legal battle with OpenAI.
The company alleges that former engineer Chang Liu accessed confidential Apple hardware information after joining OpenAI, used that information in AI-assisted engineering work and was involved in conduct that may have resulted in evidence being destroyed.
OpenAI disputes Apple's broader case and has argued that Apple's own security practices undermine its trade-secret claims.
The court will ultimately have to separate allegation from evidence.
But regardless of who wins, the dispute exposes a much bigger problem facing the technology industry.
The AI talent war is becoming a data war.
Engineers carry knowledge.
AI systems can process that knowledge.
Cloud platforms can preserve it.
And increasingly sophisticated forensic tools can reconstruct what employees did with it.
That means the most valuable corporate secrets of the next decade may not be locked inside a company's headquarters.
They may be sitting inside an engineer's laptop, cloud account—or AI agent.
And Apple vs. OpenAI may be one of the earliest major legal battles over who gets to control that information when the people who created it walk out the door.
Comments
Post a Comment