Satya Nadella says businesses adopting AI must retain control of the knowledge and intellectual property they create with it. The Microsoft chief executive also argues that people will increasingly steer and check AI agents, while stronger models bring new security and governance challenges.
Microsoft AI Watch analysis
What happened
Nadella made the remarks in a conversation with Commonwealth Bank chief executive Matt Comyn at the bank’s Accelerate AI conference in Sydney. He said companies should avoid letting their knowledge become locked inside a particular model, and should build their ability to learn and use that knowledge into a competitive advantage.
The conversation focused on three connected concerns: keeping control of intellectual property, redesigning work around people and AI agents, and strengthening governance and cyber defences as AI capabilities grow. CommBank published the account on 18 November in its conference report.
Why it matters
This is a practical question for companies, not just a debate over which model tops a leaderboard. If useful knowledge and feedback accumulate inside AI services, businesses need to consider how they retain access and control when their tools or suppliers change. And if agents take on more work, someone still needs to set their boundaries and check what they do.
Nadella’s remarks are a view from a company building AI products, rather than a neutral forecast. But the tension he identifies is real enough to deserve attention: adopting a capable system can create value while also making a company more dependent on it.
Our read
The interesting challenge is to treat ownership, oversight and security as part of AI adoption from the start, not as paperwork to fetch after the agent has taken the scenic route through the company’s files. “Keep control” is a useful principle; businesses will need to turn it into concrete rules for data, portability and human responsibility.
What to watch
- Whether Microsoft offers more specific guidance on keeping company knowledge portable across AI models.
- How businesses define which work agents may perform and who checks the result.
- Whether AI governance plans address cyber risk alongside intellectual-property and data controls.
Discussion spark: Should businesses be required to keep their AI-generated knowledge portable between providers, or is that a commercial choice each company should make for itself?
Sources and evidence
- A conversation between CBA's Matt Comyn and Microsoft's Satya Nadella (18 November 2026, 00:00 UTC)
not affiliated with or endorsed by Microsoft