Legal Pressure Mounts on Altman as OpenAI Uncovers Hacks
For OpenAI, security breaches and courtroom errors are converging into a broader test of trust in artificial intelligence. The company has uncovered dozens o...
For OpenAI, security breaches and courtroom errors are converging into a broader test of trust in artificial intelligence. The company has uncovered dozens of hacks, while cases involving AI-generated mistakes are adding pressure on lawyers and technology firms to show they can use the tools responsibly.
The available reporting, however, does not provide details about the hacks, when they occurred or whether they led to legal action against OpenAI chief executive Sam Altman. The headline-level claim points to mounting legal risks, but the material provided does not establish what those risks are or connect them directly to the breaches.
AI errors are reaching court
Separately, lawyers in the United States have faced scrutiny over filings containing fabricated testimony or case citations generated with AI. In three cases reported within a three-day span, a defense lawyer was rebuked over invented police-witness testimony in a New Mexico murder case; prosecutors in Oklahoma disclosed that a judge’s order had cited nonexistent cases; and a lawyer for State Farm in Los Angeles was fined over inaccurate citations in seven filings.
The incidents add to a pattern of AI-generated errors appearing in court documents and other legal work. A Manhattan judge who sanctioned lawyers in 2023 for citing fake ChatGPT cases said he expected the profession to be more alert to the risk of fabricated output.
That concern is shaping how law firms train staff. Firms are pairing instruction on how to use AI systems with guidance on when not to rely on them, and on who remains responsible for checking their output. “The rules haven’t really changed,” legal ethics expert Lucian Pera said, but convincing people to follow them can be difficult when AI responses sound persuasive.
Security and accountability
The two issues raise different questions: whether systems and company networks can be protected from intrusion, and whether people using AI can verify its answers before acting on them. The supplied material does not explain how OpenAI’s reported hacks relate to the court cases, or whether the breaches exposed user information.
Law firms are making ethical and governance training part of broader AI instruction. Morgan Lewis describes its approach as a “license to drive” for using AI in legal services, while stressing that responsible use should be woven into professional training rather than treated as a one-off lesson.
Further details about the OpenAI incidents — including their scope and any resulting legal proceedings — will determine how directly the reported hacks translate into exposure for Altman and the company.
Source: ft.com
