Microsoft’s Nadella to businesses: embracing AI can drive you bankrupt
Don’t share too much information with third-party AI models.
Image by Cybernews.
- Microsoft CEO Satya Nadella warned businesses not to overtrust third-party AI or share too much proprietary data.
- According to Nadella, companies should keep AI usage data to train or fine-tune their own open-weight models.
- He argued firms need multiple AI models so one vendor cannot control their operations or costs.
- Cybersecurity expert Ilia Kolochenko warned companies may remain legally liable when third-party AI tools cause harm.
Key Takeaways by nexos.ai, reviewed by Cybernews staff.
Microsoft CEO Satya Nadella has previously criticized a small group of leading AI companies for dominating the conversation with either hype overload or gloom. Now, he’s warning businesses that they shouldn’t trust AI too much.
By now, businesses large and small already know that using too much AI can be extremely costly if you’re not careful with token usage.
AI token costs now exceed employee salaries at some companies, and Uber maxed out its full-year AI budget in 4 months. Tokenmaxxing is not something executives want to see anymore.
Token pricing follows a familiar trap: affordable at first to lock in dependence, then prices rise once companies are hooked. And the models by AI giants like OpenAI or Anthropic aren’t cheap.
A month ago, Nadella already spoke in an interview with The Wall Street Journal, saying AI models should be cheaper and pointing out that the public wouldn’t tolerate just a few models and companies “doing all of the learning for the world.”
Now, the Microsoft CEO has another warning. On an episode of CNN’s Fareed Zakaria GPS, Nadella said businesses shouldn’t share too much information – including proprietary data and the prompts entered – with third-party AI models they’re using.
Instead, Nadella thinks, companies should retain all data on how their staff uses AI tools so they can eventually train or fine-tune their own open-weight model.
“Any firm that doesn’t have this control, I will claim will not remain a firm because you’ve essentially outsourced your thinking,” Nadella said.
True, he might have an angle as Microsoft hosts the cloud infrastructure that businesses can run their open-weight models on.
But just so it happens that there’s been a lot of talk about open-weight models as a much cheaper alternative to closed-source models like Claude Code, for example.
When an OpenAI executive moved to complain about open-weight AI models such as Kimi K3, developed by a China-based firm, he got ridiculed all over the web. Clearly, consumers – individual users and businesses alike – still care about costs.
According to Nadella, companies should use a custom AI platform and use multiple AI models on it: “Any one model can go away, and you can still continue to be in control of your own destiny.”
If you plan to use agentic AI for security testing, you must think twice and talk to your lawyers. Otherwise, you could start getting summonses to court on a daily basis,Dr. Ilia Kolochenko
A recent incident in which OpenAI recently admitted its AI had escaped a closed environment and attacked Hugging Face on its own during a test (OpenAI later revealed the cyberattack went even further, targeting another 4 services) is also illustrative.
Dr. Ilia Kolochenko, founder of global cybersecurity company ImmuniWeb, told Cybernews that companies using a third-party AI model for security testing or any other reason should remember they’ll be fully liable if something goes wrong.
“You may then file a lawsuit against the AI vendor that you used, but your chances of succeeding in the court of law are tiny due to countless contractual disclaimers and limitations of liability that may be enforceable against you,” Kolochenko said.
“If you plan to use agentic AI for security testing, you must think twice and talk to your lawyers. Otherwise, you could start getting summonses to court on a daily basis.”
