The deal. Protectt.ai, an Indian mobile app security company backed by Bessemer Venture Partners, became the majority owner of AIShield on Tuesday. AIShield was built inside Bosch Global Software Technologies starting in 2022. Protectt.ai bought in through new shares plus a purchase from existing holders. Bosch keeps a minority stake. AIShield becomes an independent company, AI Shield Private Limited, with its own management.
Price: Undisclosed. Timeline: Undisclosed. Protectt.ai’s co-founder Manish Mimani told Inc42 the structure mixes cash and employee stock options, with more investment later as needed.
The odd part is the size gap. Protectt.ai raised around $16 million in a Series A led by Bessemer and Bajaj Finance in 2025. AIShield, per the companies, has more than 40 patent filings, 11 granted patent families, and customers across Europe and the Middle East. It also has only about 12 to 15 people, with plans to grow to 50 to 60 by March 2027.
So a small startup just took control of a big company’s research asset. That’s not as strange as it sounds.
Why corporate labs sell. Big industrial companies build security tech because they need it for their own products. Bosch makes software for cars, factories and devices, so it had real reasons to study attacks on AI models. But building a lab is different from running a security software business. Selling to banks takes a sales team, compliance paperwork and a support desk that a corporate lab rarely has.
A deal like this solves both sides. The buyer gets patents and a research team it couldn’t hire fast. The parent gets the asset off its books, keeps a slice of the upside, and doesn’t have to build a go-to-market team for a market it isn’t in.
What AIShield actually does. Three things, in plain terms. It scans AI models before launch for hidden weaknesses. It red-teams them, which means attacking them on purpose to find holes, including prompt injection, where hostile text tricks a model into ignoring its rules. And it adds runtime guardrails, filters that watch what goes in and out of a live AI app.
Why a mobile security company wants it. Protectt.ai’s core product shields banking and fintech apps on phones. It says its tech runs on more than 1 billion devices and handles 15 billion app sessions a month. Mimani’s bet is that apps will lean on AI agents so heavily that, within a year or two, many won’t need a traditional screen. If the app becomes an agent, protecting the app means protecting the agent. AIShield is the shortcut there.
Protectt.ai also set up an M&A team earlier this year, and says it’s looking at API security and cloud security next. That’s a roll-up plan, not a one-off.
The sovereignty angle. Both companies pitch this as sovereign AI security, meaning tools built and run in-country so regulated buyers keep control of their data and models. It isn’t only an Indian story. On the same day, French SOC vendor Sekoia launched a self-hosted version of its platform for government, defense and critical infrastructure customers who can’t send security data to a public cloud. Where data must stay has become a buying criterion, not a footnote.
The risks. Patents are not product-market fit. Twelve to fifteen people can’t support global enterprise accounts. Merging an AI security research platform with a mobile app shielding product is real engineering, not a press release. And with no deal value, nobody outside can tell whether this was a bargain or a rescue.
Our take. Expect more of these. Corporate labs at industrial, auto and telecom giants have spent years building AI security research nobody outside has seen. Their parents don’t want to become security vendors. Regional security startups with bank relationships want IP fast. Most of those deals will close with terms Undisclosed and draw little attention. They’ll matter anyway, because they decide which AI security tools end up inside regulated banks outside the U.S.