Apple's Shocking Evidence Against Employee Who Allegedly Stole Data for OpenAI

Apple has filed what it's calling "shocking evidence" in its case against a former employee accused of stealing sensitive company data and handing it to OpenAI. According to court documents surfaced by TechCrunch, the evidence includes allegations that the employee actively destroyed evidence of the theft after learning he was under investigation, which is a serious escalation of an already alarming case.

The case centres on a former Apple engineer who allegedly exfiltrated proprietary data related to Apple's AI and machine learning projects. The connection to OpenAI has drawn enormous attention, given the intense rivalry between Apple's own on-device AI ambitions and OpenAI's dominant position in the AI assistant market.

What Apple Says Happened

Apple's legal filing claims the employee took confidential data and, once he realised investigators were on to him, deliberately destroyed digital evidence to cover his tracks. Destroying evidence after becoming aware of an investigation is obstruction, and it substantially raises the legal stakes beyond simple trade secret theft.

The term "shocking" in Apple's filing is doing a lot of work. Legal language tends to be measured and dry, so Apple's team explicitly using that word in court documents suggests the nature or volume of the evidence genuinely surprised even Apple's internal security team. Companies don't typically use theatrical language in court filings without purpose.

What Apple allegedly found raises hard questions about its own internal security practices. If an engineer could exfiltrate sensitive AI data without triggering immediate detection, that's a gap in Apple's data loss prevention systems that will need answering. Apple is notoriously secretive and invests heavily in internal security, so the fact that this allegedly happened points to a real vulnerability.

The OpenAI Connection

The alleged link to OpenAI is the part of this case that's generating the most heat. OpenAI did not publicly comment by the time of writing. If the data was genuinely passed to OpenAI, the question becomes how much did OpenAI leadership know, and when? A rogue contractor receiving stolen data is a very different legal situation from an organisation knowingly receiving it.

It's worth noting that this case emerged during a period when OpenAI was aggressively hiring former Apple engineers, particularly those with machine learning expertise. The talent pipeline between Apple and OpenAI has been well-documented. Apple itself has been building out Apple Intelligence capabilities at speed, and the two companies are genuinely competing for the same engineers and, arguably, the same users.

What It Means for Corporate AI Security

This case is going to become a case study in corporate AI security programmes for years. The value of AI training data, model architectures, and internal benchmarking results has skyrocketed as AI has become central to corporate strategy. That makes engineers with access to that data a high-value target for competitors, state actors, and opportunists alike.

Corporations are going to have to rethink how they classify and protect AI-related intellectual property. Traditional data loss prevention tools were built for a world where the valuable asset was a customer database or a CAD file. A proprietary fine-tuned model or internal benchmark suite doesn't fit neatly into those categories.

For now, the case is proceeding through the courts. If Apple's evidence is as strong as it claims, the former employee is in a very difficult position. The destruction of evidence allegation alone could result in sanctions that go well beyond whatever the original data theft charge carries. This story is not done by a long way.