This week, OpenAI quietly published a statement on its website pledging not to use its patents offensively.
OpenAI emphasized its commitment to the principles of “broad access” and “collaboration” and said it would only use patents in ways that support innovation. “We will not engage in any activity where a party threatens, asserts a claim, files a lawsuit, assists anyone else in such activity against us, or harms us or our users. Unless otherwise involved, we pledge to use our patents only defensively.” company.
But experts say that claim is just whitewashing.
In contrast to patent pledges, such as IBM's 2005 promise not to use 500 specific patents for open source software makers, OpenAI's pledge is vague and ambiguous, according to MBHB partner Mike Borella. It is said that It's not clear what exactly “protection” means, or what activities OpenAI would recognize as “harmful” to a company or its users.
“That last phrase seems like an exception that ignores the rule,” Borella told TechCrunch. “This can be read in a variety of ways, including covering all of OpenAI’s competitors in the market and parties that fairly criticize ChatGPT’s flaws.”
He also argued that OpenAI doesn't have much of a patent portfolio that it could claim if it wanted to. Its IP strategy relies heavily on trade secrets, including confidential training data and methodologies. Patent applications with the U.S. Patent and Trademark Office are typically not published until 18 months after filing, but only a few OpenAI applications have been approved to date.
Additionally, there's the fact that OpenAI's patent pledge is just that: a pledge. Like contracts and warranties, they are non-binding and voluntary.
Shubha Ghosh, a Syracuse University professor who specializes in intellectual property law, said it's not clear what legal weight, if any, it would have. In any case, he noted, this does not limit OpenAI from exercising its rights under copyright law, trade secret law, and contract law.
“The computer source code that guides ChatGPT will be protected by copyright and trade secret laws,” Ghosh said. “The provisions that OpenAI includes in its Terms of Service remain contractually enforceable. If OpenAI takes any action that deviates from its commitments, such as suing other companies or individuals for patent infringement, OpenAI may temporarily lose is reliability.”
Borella called OpenAI's statement “a demonstration of the virtues of public relations.” This is an attempt to curry favor with the technology community and regulators in line with Tesla's vaguely defined patent pledge made in 2014.
“The statement as a whole is more of a papier-mâché tiger than a substantive attempt to foster healthy competition in the marketplace,” Borella said.