Microsoft knew about data theft for AI, according to lawsuit

The generative artificial intelligence industry is facing growing legal pressure over the use of protected data. A lawsuit filed by The New York Times against Microsoft and OpenAI has brought to light internal documents showing that senior executives and employees of both companies were aware that they were using others' intellectual property without permission to train their language models.

According to the declassified court documents, as early as January 2023, Brent Hecht, Microsoft's director of applied science, described in an internal memo the use of New York Times articles without consent as "an astonishing theft of unprecedented proportions" and "the greatest theft of work in human history." These warnings came nearly a year before the newspaper filed its lawsuit in December 2023.

Warnings ignored from within

The documents reveal that this was not an isolated case. In June 2023, Nick Turley, the OpenAI executive responsible for the development of ChatGPT, stated that AI represented an "existential threat" to publishers. Even earlier, in 2020, Jack Clark, OpenAI's policy director, warned in a memo addressed to Sam Altman and Greg Brockman that the company's work "will lead us increasingly to create systems that replace the work of the people who define society's culture."

These internal warnings did not stop development. The lawsuit claims that both ChatGPT and Copilot have generated verbatim copies of Times articles and produced falsehoods attributed to the newspaper, which has contributed to the loss of subscribers and advertising revenue.

Implications for SMEs

This case highlights an uncomfortable reality: the generative AI tools that many companies use daily have been built, in part, on data obtained without authorization. For an SME, this poses risks that go beyond ethics:

• Indirect legal risk: If your company uses AI-generated content that reproduces protected material, you could face copyright claims.
• Dependence on questioned providers: Entrusting critical processes to platforms whose legal basis is in question can affect business continuity.
• Reputation: Customers increasingly value transparency in the use of AI. Ignoring the origin of data can damage brand image.

At ForgeNEX we have already analyzed how AI is redefining the workplace and what it means for your SME. This case adds an additional layer: the need to evaluate not only functionality, but also the legality and provenance of the tools we adopt.

Practical recommendations

Given this panorama, at ForgeNEX we recommend that SMEs and IT teams take proactive measures:

1. Audit the AI tools you use: Review the terms of service and privacy policies. Do they clearly indicate how the models were trained? Do they offer guarantees on data use?
2. Prioritize transparent providers: Opt for solutions that document the origin of their training data and that offer indemnities for possible intellectual property claims.
3. Train your team: Ensure that employees understand the risks of copying and pasting AI-generated content without verifying its originality.
4. Consider alternatives: Explore open-source models or solutions that allow greater control over data and its provenance.

Legal pressure on AI giants will only increase. As we pointed out in our analysis on AI in forensics, trust is not validation. SMEs must adopt a critical and documented approach before integrating these technologies into their processes.

"OpenAI will become the symbol of how Silicon Valley is thoughtlessly venturing into other areas of life and leaving a disaster in its wake," wrote Jack Clark in 2020.

This case reminds us that innovation without responsibility has a cost. For SMEs, the key is to adopt AI with eyes open, understanding both its benefits and its legal and reputational risks. At ForgeNEX we can help you evaluate which tools fit your needs and your risk tolerance.

Source: Computerworld. Analysis and adaptation: ForgeNEX.