Gartner urges businesses to 'block all AI browsers' - what's behind the dire warning

sankai/iStock/Getty Images Plus

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• AI browsers are excessively risky for adoption today, Gartner said.
• CISOs request to artifact them until existing information concerns are reduced.
• Automation is useful, but convenience shouldn't regenerate security.

Businesses should steer wide of agentic browsers for now, analysts from probe steadfast Gartner person warned. 

Agentic browsers, different known arsenic AI browsers, are changing however we usage browsers to entree the internet, execute hunt queries, and negociate workflows. 

Also: Scammers are poisoning AI hunt results to steer you consecutive into their traps - here's how

As AI has present embedded itself into astir each industry, it was lone a substance of clip earlier these technologies infiltrated the web browser market. 

Not lone are well-known browser developers exploring the benefits and imaginable applications of artificial quality successful hunt functionality, idiosyncratic assistance, support, and security, but involvement successful agentic browsers has besides led to a harvest of smaller outfits appearing, with AI browsers present being offered from organizations including OpenAI and Perplexity. 

Also: Are AI browsers worthy the information risk? Why experts are worried

For probe purposes, contented summarization, and personalization, AI browsers can beryllium beneficial. They tin prevention you clip and whitethorn summation your ratio -- but this doesn't mean that each query LLM-based chat assistant's reply is correct, nor does it mean that AI browsers are inherently secure. 

In an expert study and advisory dubbed "Cybersecurity Must Block AI Browsers for Now," published earlier this month, arsenic reported by The Register, Gartner analysts said that portion agentic browsers person the imaginable to revolutionize however we interface with websites and execute online activities, they besides present "critical cybersecurity risks."

"CISOs indispensable artifact each AI browsers successful the foreseeable aboriginal to minimize hazard exposure," the probe bureau said.

(Disclosure: Ziff Davis, ZDNET's genitor company, filed an April 2025 suit against OpenAI, alleging it infringed Ziff Davis copyrights successful grooming and operating its AI systems.)

The risks of AI browsers, explained

According to analysts Dennis Xu, Evgeny Mirolyubov, and John Watts, the main contented is that "default AI browser settings prioritize idiosyncratic acquisition implicit security."

As AI browsers are capable to run independently, interact with websites, and execute tasks connected a user's behalf, arsenic good arsenic show contented that could punctual the idiosyncratic to instrumentality circumstantial actions -- adjacent erstwhile this contented is malicious -- determination are demonstrable risks to their use, successful some a user and concern context.

Also: I've been investigating the apical AI browsers - here's which ones really impressed me

For example, an AI chatbot could unintentionally interact with a malicious website, oregon an worker mightiness taxable confidential, delicate firm information to an AI adjunct and beryllium unaware of wherever this accusation is stored. If an AI's unreality backend is not secure, this could yet effect successful a information breach oregon information incidental that places businesses astatine risk. 

In addition, Gartner said that employees could beryllium tempted to usage AI browsers oregon assistants to automate repetitive tasks. While AI tin payment workflows and amended efficiency, utilizing it successful this mode whitethorn besides person unintended effects -- specified arsenic a subordinate of unit utilizing it to implicit cybersecurity grooming portion not learning thing astatine all. 

Why it matters

Gartner is acold from the lone enactment raising the alarm astir the imaginable cybersecurity implications of AI browsers and assistants. 

Security experts are watching the AI browser improvement intimately and have warned us of galore imaginable cybersecurity challenges progressive successful regulating them, including the emergence of punctual injection attacks, idiosyncratic information theft and exposure, and the accrued hazard of surveillance that compromises our idiosyncratic security. 

Gartner recommends that businesses behaviour hazard assessments connected AI solutions arsenic good arsenic immoderate associated backend systems, arsenic specified information audits tin uncover whether oregon not AI browsers are deemed an acceptable information risk. Considering however caller and untried they are, the reply is apt no. 

Also: Use an AI browser? 5 ways to support yourself from punctual injections - earlier it's excessively late

If this is the case, Gartner said that CISOs should see blocking AI browser usage entirely. However, successful either case, employees should beryllium trained connected the dangers of relying excessively heavy connected AI -- and wherefore they shouldn't portion with delicate accusation erstwhile they are interacting with AI programs. 

"While agentic browsers committedness galore features to heighten idiosyncratic experience, we are inactive successful aboriginal stages wherever the risks are not good understood and default configurations prioritise convenience implicit security, thing we spot successful galore technologies," commented Javvad Malik, pb information consciousness advocator astatine KnowBe4. "However, broad bans are seldom sustainable semipermanent strategies. Instead, the absorption should beryllium connected hazard assessments that measure the circumstantial AI services powering these browsers. This tin let for measured adoption portion maintaining indispensable oversight."