Generative AI attacks are accelerating at an alarming rate
Two new reports from Gartner highlight the new AI-related pressures companies face, and the tools they are using to counter them
Generative AI attacks are accelerating at an alarming rate, according to Gartner, with 29% of organizations experiencing an attack on their AI application infrastructure in the last 12 months.
In a survey of 302 cybersecurity leaders in North America, EMEA, and Asia-Pacific, the consultancy found that 62% of organizations experienced a deepfake attack involving social engineering or exploiting automated processes.
Audio incidents were more common than video, with 44% reporting social engineering during a call with a supposed employee, compared with 36% in the case of video calls.
Similarly, 32% experienced deepfake audio used against automated voice biometrics, compared with 30% in the case of face biometrics or identity verification.
Analysis from the consultancy found AI assistants are now a top target for threat actors, and they’re vulnerable to a variety of adversarial prompting techniques.
Attack methods highlighted in the study included prompts aimed at manipulating large language models (LLMs) or duping multimodal models into generating malicious outputs.
All told, 32% of respondents to the Gartner survey said they’d experienced an attack of this kind over the last year, representing a significant uptick.
“As adoption accelerates, attacks leveraging GenAI for phishing, deepfakes and social engineering have become mainstream, while other threats — such as attacks on GenAI application infrastructure and prompt-based manipulations — are emerging and gaining traction," said Akif Khan, VP analyst at Gartner.
Generative AI attacks are changing the game
While 67% of cybersecurity leaders said emerging generative AI risks require significant changes to existing cybersecurity approaches, Gartner recommends a more cautious strategy.
“Rather than making sweeping changes or isolated investments, organizations should strengthen core controls and implement targeted measures for each new risk category,” said Khan.
Meanwhile, in a separate report, Gartner noted that organizations are increasingly turning to pre-emptive cybersecurity practices rather than standalone detection and response (DR).
By 2030, pre-emptive cybersecurity solutions are expected to account for 50% of IT security spending - up from less than 5% just a year ago.
Pre-emptive cybersecurity technologies use advanced AI and machine learning to anticipate and neutralize threats before they materialize. They include capabilities such as predictive threat intelligence, advanced detection, and automated moving target defense.
“Pre-emptive cybersecurity will soon be the new gold standard for every entity operating on, in, or through the various interconnected layers of the global attack surface grid (GASG),” said Carl Manion, managing vice president at Gartner.
“DR-based cybersecurity will no longer be enough to keep assets safe from AI-enabled attackers. Organizations will need to deploy additional countermeasures that act pre-emptively and independently of humans to neutralize potential attackers before they strike.”
Gartner predicts a shift from broad, one-size-fits-all DR security platforms toward more targeted pre-emptive cybersecurity tactics, many of which will be based on agentic AI and domain-specific language models (DSLMs).
Make sure to follow ITPro on Google News to keep tabs on all our latest news, analysis, and reviews.
MORE FROM ITPRO
- Security researchers have just identified what could be the first ‘AI-powered’ ransomware strain
- The top ransomware trends for businesses in 2025
- Agentic AI could be a blessing and a curse for cybersecurity
-
AI is rewriting the rules for MSPs in cybersecurityNews The inescapable presence of AI is changing what customers expect from MSPs, who are increasingly being pulled into business decisions that extend beyond technology
-
EU sues over delays to border control biometricsNews EU-LISA is suing IBM, Leonardo, and Atos over Entry/Exit System delays
-
OpenAI says its new cyber-focused AI model found thousands of Linux kernel vulnerabilitiesNews The new cyber-focused AI model boasts improved bug hunting capabilities compared to its predecessor
-
88% of UK firms have had an “agentic security incident” in the last year – ServiceNow warns more will comeNews Concerns are growing over the scale of adoption and an inability to match “machine-speed” capabilities
-
‘Rising threats and under-resourcing for cybersecurity is taking a toll on the people tasked with managing it’: Cyber teams are being pushed to breaking point – and AI is doing little to alleviate strainNews An increasingly perilous threat landscape and lack of funding means cyber teams are being stretched too thin
-
Anthropic’s Mythos AI tried to dupe devs in social engineering attack, collaborated with other agentsInter-agent collaboration is a serious cause for concern, says security expert
-
Anthropic joins OpenAI in admitting loss of control in cybersecurity testsThe company found Claude AI had escaped containment three times and targeted other organizations
-
'It delivers world-class performance at 50 percent of the cost of leading models': Microsoft unveils cut-price AI for security with latest in-house model launchNews Pairing the MAI security model with GPT-5.4 gives benchmark leading results at half the cost, according to the tech giant
-
The case for the channel in an AI-driven security marketIndustry Insights AI won't replace channel partners; SMB cybersecurity still relies on trust
-
Cisco just launched two cyber-focused small language models: Antares-350M and Antares-1B aim to supercharge codebase analysis – and they run at a “fraction of the compute expense” of popular frontier modelsNews The Antares models unveiled by Cisco aim to cut costs in codebase analysis
