'You need your own bots' to wage war against rogue AI, warns Varonis VP
Infosec pros are urged to get serious about data access control and automation to thwart AI breaches


The world is facing an AI hurricane, and organizations must batten down the hatches by securing their data vaults and deploying "bots" to combat rogue AI.
So warns, Matt Radolec, vice president of Incident Response and Cloud Operations at Varonis, who said that AI poses an existential threat to their organizations if they fail to control data access and police AI prompts.
"AI is the biggest opportunity and biggest threat to your organization," Radolec declared, setting the tone for his RSA Conference 2024 talk "Reducing AI's Blast Radius: How to Prevent Your First AI Breach."
Radolec, who has spent over 15 years safeguarding sensitive data from state secrets to corporate jewels, argued that the obsession with malware, threat actors, and CVEs has distracted organizations from the real prize: data. "Data is where the damage happens. Data is where you'll feel the pain of AI," he cautioned.
Drawing on real-world examples from Varonis' incident response investigations, Radolec highlighted the grave consequences of data breaches and corruption,. These were wide-ranging and included disrupting Alzheimer's research, crippling a city's utilities, all the way through to causing a literal "sh*tstorm" by compromising sewer systems.
"We all know an AI superstar when we see one, and Jensen Huang nailed it. AI is a data problem," Radolec said, quoting the Nvidia CEO. "Your data is your company's source code. It's intellectual property. It's worth a lot."
To combat the AI tempest, Radolec urged organizations to shift their focus from endpoints to data vaults, monitoring every transaction, detecting anomalies, and policing every AI prompt. He stressed the importance of granular access control, noting that the average organization has 17 million files open to all employees and over 40 million unique access control lists to manage.
Get the ITPro daily newsletter
Sign up today and you will receive a free copy of our Future Focus 2025 report - the leading guidance on AI, cybersecurity and other IT challenges as per 700+ senior executives
"You have to police your prompts. Think about it. Has anyone ever gotten a speeding ticket or another type of moving violation? I know I have a few. Does the fear of getting one of those stops you from driving like a maniac? You have to issue tickets and take reckless drivers off the road even more so when people abuse their co-pilots," Radolec advised. “Because having weak access controls and not policing your prompts would be akin to giving every employee a Ferrari and letting them loose to race on residential streets.”
RELATED WHITEPAPER
Perhaps his most provocative suggestion was the need for organizations to deploy AI and automation to combat rogue AI. "If you want to survive AI, you will need your own bots on your side. Automation and AI is the only way to combat AI. Trust me," he said, leaving the audience to ponder the impending bot wars.
Radolec concluded his talk by urging attendees to embrace their role as data protectors, reminding them: "Data is looking up at you and it's saying 'Help me RSA conference attendees. You're my only hope.'"
Rene Millman is a freelance writer and broadcaster who covers cybersecurity, AI, IoT, and the cloud. He also works as a contributing analyst at GigaOm and has previously worked as an analyst for Gartner covering the infrastructure market. He has made numerous television appearances to give his views and expertise on technology trends and companies that affect and shape our lives. You can follow Rene Millman on Twitter.
-
RSAC Conference 2025: AI and quantum complicate security
Organizations are grappling with the complications of adopting AI for security
-
RSAC Conference 2025 was a sobering reminder of the challenges facing cybersecurity professionals
Analysis Despite widespread optimism on how AI can help those in cybersecurity, it’s clear that the threat landscape is more complex than ever
-
RSAC Conference day three: using AI to do more with less and facing new attack techniques
-
"There needs to be an order of magnitude more effort": AI security experts call for focused evaluation of frontier models and agentic systems
News Evaluating the risks of dynamic, evolving AI networks is slow work for cybersecurity analysts
-
Cyber defenders need to remember their adversaries are human, says Trellix research head
There's a growing overlap between nation-state actors and cybercriminals, but these attackers are real people who make mistakes
-
RSAC Conference day two: A focus on what attackers are doing
From quantum to AI, experts discussed how new and experimental technologies could be used by hackers to access and decrypt sensitive data
-
RSAC Conference Day One: Vibe Is 'All In' on AI for Security
News Artificial intelligence took center stage as RSAC Conference looks at how the discussion has moved from generative AI to agentic AI
-
RSAC Conference 2025 live: All the latest from day three
Live blog ITPro is covering RSAC Conference 2025 live – find out all the day-three news right here