A new N-able survey of 595 cybersecurity professionals finds that AI security adoption is moving faster than organizational trust and governance. About 86.1% of respondents saw security alert volumes rise over the past year, and more than 80% experienced a confirmed or suspected AI-assisted attack. Despite increased adoption of AI-powered tools, only 24.5% fully trust autonomous AI remediation without human review.
Brendan Griffin, Director of Threat Research at N-able, noted that AI can be non-deterministic, which makes it hard to interpret without additional context. The findings suggest that simply adding more AI tools does not automatically improve security outcomes. Organizations are reportedly making exceptions to security policies to enable these new systems.
Separately, Anthropic launched Cyber Mission, a project that provides AI models and threat research aimed at securing critical infrastructure. The free tools sparked debate on Reddit about whether they could eliminate entry-level security jobs. Experts worry that newcomers may lose career ladders even as AI removes routine tasks from the field.
Intel argued at the Open Compute Project Global Summit in San Jose that the AI gold rush is shifting away from large models in centralized data centers. The company is promoting prompt routing across open hardware platforms, signaling a broader move toward more distributed and open AI infrastructure.
On the social side, sociologist Sherry Turkle told The Ezra Klein Show that AI and communication technologies are deskilling our ability to deal with one another. She discussed how tools from social media to short-form content shape how we think and see each other, raising questions about the long-term human impact of these systems.
MIT researchers left an AI agent named Eko alone on a virtual island for up to 30 hours with no instructions. Eko explored independently, built structures, spelled its name with blocks, and created drawings like mandalas. The study points to a possible natural drive in AI to play and explore, similar to how young children learn through interaction.
Federal agencies may soon get a clearer picture of employee concerns about AI. A proposed overhaul of federal employee surveys, directed by the Office of Personnel Management, would ask workers about job security, approved use, verification of AI outputs, and their ability to raise concerns. Experts say the results should guide smarter AI investment while protecting public service standards.
University of Pennsylvania researchers used AI to analyze nearly 150,000 Reddit posts about retatrutide, an unapproved GLP-1 weight-loss drug. The tools identified about 14,000 users who reported taking the gray-market version and extracted reported symptoms. The most common self-reported symptom was increased appetite, which differs from clinical trial side effects, though researchers stress the findings are a starting point for hypotheses.
School principals are turning to AI tools to tackle problems like absenteeism. Derek Cantrell of Allegheny High School built an app that tracks student absences and rewards students with points for improvement. Other principals use AI to help teachers write yearly goals and track student progress, though experts warn about keeping student data safe and localized.
Writer Wendell Berry's ideas offer a philosophical challenge to how we live with AI. Berry argued that the next great division in the world will be between people who wish to live as creatures and those who wish to live as machines. He believed democracy requires inefficient but more equal processes rather than pure automation, even as AI promises convenience.
David Lindner, chief information security and data privacy officer at Contrast Security, recommends that organizations start with the CISA Known Exploited Vulnerabilities catalog and use the Exploit Prediction Scoring System. He argues that AI agents should sort and evaluate findings rather than simply generating more of them, while CISOs examine detection and containment controls at the application layer.
Key Takeaways
- N-able survey of 595 cybersecurity pros finds 86.1% saw alert volumes rise, but only 24.5% fully trust autonomous AI remediation.
- Over 80% of organizations experienced a confirmed or suspected AI-assisted attack in the past year.
- Anthropic launched Cyber Mission, sparking debate over whether free AI tools could eliminate entry-level security jobs.
- Intel promotes prompt routing on open hardware, arguing the AI gold rush is shifting away from centralized large models.
- MIT's Eko AI agent explored a virtual island for 30 hours, building structures and drawings without instructions.
- Federal employee surveys are being overhauled to capture worker concerns about AI, including job security and output verification.
- UPenn researchers used AI to analyze 150,000 Reddit posts on unapproved GLP-1 drug retatrutide, finding increased appetite as a common self-reported symptom.
- School principals use AI apps to track absenteeism and help teachers set goals, with warnings to keep student data localized.
- Contrast Security's David Lindner urges CISOs to start with CISA Known Exploited Vulnerabilities catalog when using AI vulnerability scanners.
- Sherry Turkle and Wendell Berry raise broader concerns that AI may deskill human interaction and push people toward living as machines.
N-able Study Reveals AI Security Faces Trust and Complexity Barriers
A global survey of 595 cybersecurity professionals in the US and UK found that AI security adoption is outpacing trust and governance. 86.1% of respondents saw security alert volumes increase over the past 12 months. Only 24.5% fully trust autonomous AI remediation without human review. The study highlights that adding more AI tools does not automatically produce better security outcomes.
N-able Research Shows AI Security Adoption Outpaces Organizational Trust
New research from N-able finds the security industry is at an inflection point as Frontier AI models transform cyber defense and attacker capabilities. 81.0% of organizations increased AI-powered security tool adoption in the past year, yet 75.5% still do not fully trust AI remediation. Brendan Griffin, Director of Threat Research, noted that organizations need AI that reduces operational complexity and delivers meaningful context.
AI Security Tools Trigger Alert Fatigue in Organizations
According to N-able, over 86% of respondents reported increased security alert volumes in the past 12 months, with nearly a quarter seeing a surge of over 50%. More than 4 in 5 experienced a confirmed or suspected AI-assisted attack. Brendan Griffin explained that AI can be non-deterministic, making it challenging to interpret without additional context. Organizations are making exceptions to security policies to enable AI tools.
Experts Urge Smarter Prioritization for AI Vulnerability Scanners
David Lindner, chief information security and data privacy officer at Contrast Security, recommends starting with the CISA Known Exploited Vulnerabilities catalog and using the Exploit Prediction Scoring System. He emphasizes that AI agents should sort and evaluate findings rather than simply generating more of them. Lindner also urges CISOs to examine detection and containment controls, particularly at the application layer, while remediation continues.
AI Is Making Us Antisocial, Says Sherry Turkle
On The Ezra Klein Show, sociologist Sherry Turkle explains how AI and communication technologies are deskilling our ability to deal with one another. She discusses how tools from social media to short-form content shape how we think and see each other. The video opinion piece explores the social impact of artificial intelligence on human interaction.
Anthropic's Cyber Mission sparks debate over AI and junior security roles
Anthropic launched Cyber Mission to provide AI models and threat research for securing critical infrastructure. Security professionals on Reddit debate whether the free tools could eliminate entry-level jobs. Experts worry that newcomers may lose career ladders even as AI removes routine tasks. The discussion highlights how technology progress may not benefit all workers equally.
Intel says AI gold rush is over and promotes prompt routing on open hardware
Intel spoke at the Open Compute Project Global Summit in San Jose about the future of AI. The company argued that AI is shifting away from large models in centralized data centers. Intel is pushing prompt routing across open hardware platforms. This signals a move toward more distributed and open AI infrastructure.
Federal AI adoption surveys overlook employee concerns and human factors
A proposed overhaul of federal employee surveys could help agencies understand hesitations about AI. The Office of Personnel Management directed agencies to ask workers about their concerns regarding AI tools. Questions would cover job security, approved use, verification of AI outputs, and ability to raise concerns. Experts say the survey results should guide smarter AI investment while protecting public service standards.
Wendell Berry's vision challenges how we live with AI as creatures not machines
A new article explores writer Wendell Berry's ideas about living as ecological beings in an age of AI. Berry argued that the next great division in the world will be between people who wish to live as creatures and those who wish to live as machines. The piece examines how AI promises convenience but may ignore long-term consequences. Berry believed that democracy requires inefficient but more equal processes rather than pure automation.
Researchers use AI and Reddit to study side effects of unapproved GLP-1 drug
University of Pennsylvania researchers analyzed nearly 150,000 Reddit posts about retatrutide, an unapproved GLP-1 weight-loss drug. The AI tools identified about 14,000 users who reported taking the gray-market version and extracted reported symptoms. The most common self-reported symptom was increased appetite, which differs from clinical trial side effects. Researchers stress the findings are a starting point for hypotheses, not proof that the drug causes those symptoms.
Principals Use AI to Tackle School Challenges Like Absenteeism
School principals are using AI tools to address major problems in schools. Derek Cantrell of Allegheny High School built an app that tracks student absences and rewards students with points for improving. Other principals use AI to help teachers write yearly goals and track student progress. Experts warn principals to keep student data safe and localized.
MIT Researchers Leave AI Agent Alone on Virtual Island
MIT researchers left an AI agent named Eko on a virtual island for up to 30 hours with no instructions. Eko explored on its own, built structures, spelled its name with blocks, and created drawings like mandalas. The study shows AI may have a natural drive to play and explore, similar to how young children learn through interaction.
Sources
- N-able Research Finds AI Security Tools are Threatened by Trust and Complexity Challenges
- N-able Research Finds AI Security Tools are Threatened by Trust and Complexity Challenges
- AI security tools trigger alert fatigue
- AI Scanners Demand Smarter Vulnerability Prioritization
- Video: Opinion | A.I. Is Making Us Antisocial
- Anthropic cybersecurity AI divides security pros on Reddit
- Intel says AI 'gold rush' is over, pushes prompt routing across open hardware
- Federal AI adoption metrics miss the human factor
- Are you a creature or a machine? How to live in ecological community in a time of AI
- Engineers use AI, Reddit to identify potential side effects of unapproved new GLP-1
- Can AI Crack the Code on Some of the Biggest Challenges in Schools?
- Researchers Stranded an AI Agent on a Virtual Island. Things Got Very Weird.
Comments
Please log in to post a comment.