Artificial intelligence is getting smarter, faster and more autonomous. According to Seth Boro, so are the risks.
Boro, a Managing Partner at Thoma Bravo, told Bloomberg TV on October 19, 2026 that AI development depends on proper cybersecurity, especially for consumers. His central warning was that threats from “rogue agents” will persist as AI builds out.
That message carries extra weight given where it came from. Thoma Bravo is one of the biggest investors in cybersecurity, and its portfolio has a direct financial stake in whether people trust AI enough to use it.
Rogue agents and the new social engineering
Boro’s main concern was rogue agents. In this context, that means AI systems that behave maliciously, whether they are built that way or turned to bad ends.
Boro argued that these threats are not a passing phase. In his view, they will likely grow alongside AI deployment. More agents running in more places means more opportunities for something to go wrong.
He also drew a comparison that anyone in security will recognize. Boro pointed to prompt engineering as a vulnerability similar to human social engineering.
Social engineering is the classic con. A scammer calls an employee, pretends to be IT, and talks their way into a password. Prompt manipulation applies the same trick to software. An attacker crafts inputs that persuade an AI model to ignore its rules or leak data.
Boro said Thoma Bravo’s portfolio companies are addressing these AI threats, including building defenses against prompt manipulation. The firm’s companies are also collaborating on protections against AI-driven attacks, according to the Bloomberg interview.
For everyday users, his advice was caution.
A portfolio built for this moment
Thoma Bravo has spent years assembling a large cybersecurity business. Its portfolio generates approximately $8 billion in cybersecurity-related revenue.
Boro has been at the center of that effort. He joined the firm in 2005 and has led its cybersecurity strategy since. Thoma Bravo has backed dozens of cybersecurity companies since 2008 and has completed more than 15 exits in the sector.
The firm has stayed active recently. Thoma Bravo announced a $5.3 billion acquisition of Darktrace on October 9, 2026. Darktrace is known for using machine learning to spot unusual activity inside networks.
Then, on the same day as Boro’s interview, Proofpoint agreed to acquire Varonis for $6.3 billion. Proofpoint focuses on email and people-centric security, while Varonis specializes in protecting data.
Proofpoint’s own numbers help explain the urgency. The company processes trillions of messages and has reported a doubling of advanced threats year-over-year.
What this means
For businesses deploying AI agents, the practical takeaway from Boro’s comments is that security cannot be bolted on afterward. If AI adoption depends on trust, as he suggests, then companies rolling out agents without strong guardrails are taking on risk that may not show up until something breaks.
For consumers, the stakes are more personal. An AI assistant with access to your inbox and accounts is useful right up until someone convinces it to work for them instead.
What to watch next: how the Darktrace and Varonis deals are integrated, whether more cybersecurity consolidation follows, and how quickly defenses against prompt manipulation mature.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.