When Should an AI Company Hire Its First CISO?

The short answer is usually when its first enterprise customers start demanding formal security reviews, when it enters regulated markets, or when its AI agents gain access to sensitive business systems. At that point, the people building the product are too close to it to oversee its security on their own.

Cohere named Jaya Baloo its first chief information security officer on November 12, weeks after moving to acquire AISLE, the AI cybersecurity company she co-founded. For AI companies selling to governments and regulated enterprises, the appointment raises a hiring question: how long should engineering leaders own security before the business needs a dedicated CISO?

Quick Read

  • Cohere hired Jaya Baloo as its first CISO weeks after agreeing to buy her company, AISLE.
  • Anthropic, OpenAI, Cognition and Backstory have each set up the role differently, from an early technical hire to one executive who also builds security products.
  • Every one of the 32 CISO and CSO searches Christian & Timbers tracked in the first seven months of 2026 listed agentic AI expertise as the top requirement.
  • Boards should settle which risks the CISO owns, and who the CISO reports to, before the search begins.

Why Cohere Brought Security Leadership In-House

The move started before Baloo's formal appointment. On October 15, Cohere announced plans to acquire AISLE, putting AI bug-hunting next to the team building its models. According to Cohere, AISLE identified more than 500 previously unknown vulnerabilities across more than 100 codebases in its first year out of stealth.

Cohere also split the leadership roles. Alongside Baloo's appointment, AISLE co-founder Ondrej Vlcek became senior vice president of Cohere Cyber, a new cybersecurity business, and Stanislav Fort became vice president of AI security. Company-wide security strategy sits with one executive, while customer-facing cybersecurity products and AI security research sit with others.

Joelle Pineau, Cohere's chief AI officer, put it this way on LinkedIn: "in the area of agentic AI, cybersecurity is becoming the defining challenge of enterprise AI."

A security engineering team makes applications safe to deploy. Someone else has to assess risk across the whole company and challenge product decisions when needed, and as deployments grow, one engineering leader has a harder time doing both.

How Other AI Companies Have Timed the Hire

Anthropic moved early. Jason Clinton joined as its first CISO in 2023 after more than a decade at Google, where he led Chrome infrastructure security. Anthropic later made Vitaly Gudanets its CISO. He had spent nearly four years as Netflix's CISO and earlier served as a vice president of security engineering at Google. Clinton stayed on as deputy CISO. The first hire was a technical security leader. The second brought experience running security at a large consumer company.

OpenAI divided the work another way. In October 2024 it hired Dane Stuckey, Palantir's CISO for more than six years, to serve alongside head of security Matt Knight. Stuckey said at the time that "security is germane to OpenAI's mission."

Cognition, the company behind the coding agent Devin, made the opposite call from Cohere. In September, it brought in Alex Stamos as chief security officer, with a mandate to tighten the company's own security and to build new security products. Cohere split those jobs between executives, and Cognition handed both to Stamos.

Smaller AI companies are making the hire as well. In May, Backstory, the AI platform for revenue teams formerly known as People.ai, named former Meta security leader Victor Chang as its first CISO.

[Table 1]

Timing is the first difference. Anthropic put a technical CISO in place early and added an enterprise CISO later, while Cohere appointed its first CISO as its enterprise AI agents were gaining access to customer systems.

Scope is the other. Cognition's security chief protects the company and builds products, while Cohere's protects the company and leaves product work to Cohere Cyber. Neither model is wrong, but a company has to pick one before it writes the job description.

When Does an AI Company Need Its First CISO?

No headcount or revenue number settles it. What matters is whether the people building the product still check their own security work.

An AI startup with a handful of customers might keep security under its CTO. That gets harder as enterprise customers start demanding formal security reviews and as agents reach deeper into business systems.

Cohere had reached that point. North 2 connects AI agents to enterprise applications such as Slack and SharePoint and lets customers run the technology inside their own infrastructure. Each new integration raises questions about who sees what information and what an autonomous system is allowed to do.

A CTO still builds those controls. The board needs someone else to judge whether they work, spot risks across the business and raise concerns when a product decision puts the company at too much risk.

Wait too long and the new CISO inherits every decision made before they arrived. That's a hard way to start the job.

What Cohere's Choice Says About the Next CISO

Baloo spent years leading security at established technology companies before helping launch AISLE. She brings experience running enterprise security programs and building AI-based cybersecurity tools.

Cohere also gained dedicated AI security research leadership through the acquisition. That points to a model other AI companies might follow: the CISO owns company-wide security strategy, while specialist technical leaders secure the models and build defensive technology.

That combination of experience is hard to find. Christian & Timbers reports that all 32 CISO and CSO searches it tracked from January to July 2026 named deep agentic AI expertise as the most critical qualification, up from nine of 22 a year earlier.

For a CEO hiring a first CISO, that leaves a real choice. Should the search favor someone who has managed security across a large enterprise, or a technical leader who understands how AI systems behave in production?

The answer depends on the job. A company building foundation models needs a different security leader from one deploying third-party agents across internal systems. Boards should decide which risks the role will own before setting candidate requirements.

Who Should the First CISO Report To?

Cohere's announcement describes Baloo's responsibilities but doesn't specify her reporting line. For other AI companies planning a similar hire, that's worth deciding early.

A CISO who reports to the CTO has security concerns competing with engineering delivery. A direct line to the CEO gives the role more independence, though access to senior leadership does little without authority to influence decisions.

The pressure behind that decision is growing. BCG's 2026 CISO Survey found that 35% of organizations had experienced significant impact from AI-enabled attacks in the previous 12 months. BCG advises CEOs to set clear decision rights among the CISO, the chief risk officer and the CIO or CTO.

Start the Search Before the Next Launch

For AI businesses moving into regulated industries, the first CISO search should start well before the next major product launch. By the time a customer's security review asks who owns AI risk, that answer should already have a name.

Frequently Asked Questions About Hiring a First CISO

What's the Difference Between a CISO and a CSO?

At most companies, not much. Cybersecurity Dive notes the two titles are now seen as largely interchangeable. The traditional split puts information systems and data under the CISO, while the CSO also covers physical security and employee safety. CSO Online adds that smaller companies often have one security executive, called a CSO, with IT security reporting to them. Cognition gave Alex Stamos the CSO title for a job that covers protecting the company and building security products.

Is a Fractional CISO Enough for an Early-Stage AI Company?

For a company with a few customers and no agents inside customer systems, often yes. A fractional CISO is a part-time executive who sets security strategy and prepares the company for its first audits without a full-time salary. The case for part-time gets weaker once agents connect to customer systems the way Cohere’s North 2 does, because security questions then come up in nearly every product decision.

What Does a First CISO Cost?

More than most founders budget for. In IANS research, CISOs at companies with up to $1 billion in revenue averaged $415,000 in total compensation, and tech-sector CISOs averaged $844,000. An AI company competing with Anthropic and OpenAI for senior security talent should understand the higher end of the technology market before setting its compensation range. Equity is part of compensation for 70% of CISOs surveyed by IANS and Artico Search, and can account for up to half of total pay among top earners.

Should the First Security Hire Be a CISO or a Security Engineer?

That depends on what's going wrong. If deals stall because nobody owns security risk or has authority to answer for it, adding another engineer may leave the underlying problem unresolved. Tas Jalali, head of cybersecurity at AC Transit, argued in an EC-Council article that startups that hire engineers first end up fixing surface problems while bigger risks go unowned. He recommends a fractional CISO before Series A and adding engineers as the company grows. Anthropic got some of both by putting a technical leader, Jason Clinton, in the CISO seat.

[Table 1] Company | Security Leader | How the Role Is Set Up Anthropic | Jason Clinton as first CISO, later Vitaly Gudanets, former Netflix CISO | Technical CISO first, then an enterprise CISO, with Clinton as deputy Backstory | Victor Chang, former Meta security leader, as first CISO | First CISO at a smaller AI company Cognition | Alex Stamos as chief security officer | One executive protects the company and builds security products Cohere | Jaya Baloo, AISLE co-founder, as first CISO | CISO owns company-wide security, with products and AI research under other leaders OpenAI | Dane Stuckey, Palantir's CISO for more than six years | CISO works alongside head of security Matt Knight

Recent Articles