{"slug": "corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc", "title": "Corporate bureaucracy could give AI attackers an advantage over defenders, NCSC warns", "summary": "Britain's National Cyber Security Centre warned in a blog published by Dave Chismon, its CTO for Architecture, that organisational bureaucracy will give AI attackers an advantage over defenders, since attackers face mainly technical hurdles while defenders must navigate budgets, policies, approvals and the risk of disrupting the systems they protect. Chismon cited researcher Halvar Flake's maxim that \"All offensive problems are technical problems, and all defensive problems are political problems,\" and proposed a framework for estimating the risk of autonomous defensive actions based on five factors: potency, scope, criticality, rollout confidence and recoverability.", "body_md": "[Security](https://www.machine.news/tag/security/)\n\n# Corporate bureaucracy could give AI attackers an advantage over defenders, NCSC warns\n\n“All offensive problems are technical problems, and all defensive problems are political problems.”\n\nAttackers will gain an advantage over good guys in the age of autonomous [AI](https://www.machine.news/tag/ai/) because organisational \"politics\" will make it harder to deploy the same technology defensively, Britain’s National Cyber Security Centre has warned.\n\nAs attackers use increasingly capable AI agents, defenders face a problem technology alone cannot solve: their own bureaucracy.\n\nIn a new blog published today, Dave Chismon, the NCSC’s CTO for Architecture, argues that attackers are largely constrained by technical problems, while defenders must navigate budgets, policies, approvals and the risk of breaking the systems they are trying to protect.\n\nHe begins with a [security](https://www.machine.news/tag/security/) maxim from researcher Halvar Flake, who said: “All offensive problems are technical problems, and all defensive problems are political problems.”\n\nAttackers need to answer relatively straightforward technical questions such as: Do I have an exploit for a vulnerability? How can I avoid detection?\n\nDefenders face another layer of problems, not to mention floors of tutting middle managers.\n\nCan they get the budget to replace an insecure system? Will IT operations make time for patching? Can they get a change request for new firewall rules approved?\n\n“In other words, defenders are most restricted by their organisational policies, whilst most attackers are restricted by technical hurdles,” Chismon wrote.\n\n## Slowness-as-a-service: The perils of corporate \"safety\"\n\nDefensive problems are harder to reduce to a simple objective, because they are not \"mostly technical\" and don't always have a \"clear success state\", Chismon wrote.\n\nAn attacker can give an AI agent a technical objective and let it pursue that goal. A company considering giving an AI system control over its own infrastructure faces a much more complicated decision made in partnership with other members of their team (some of whom don't have the foggiest about security).\n\nDefensive actions also have the potential to disrupt the systems they are supposed to protect.\n\n## READ MORE: [AI loss of control is already “in the rearview mirror,” says MIT professor Max Tegmark](https://www.machine.news/ai-loss-of-control-is-already-in-the-rearview-mirror-says-mit-professor-max-tegmark/)\n\nA patch could take down a VPN, for example. Maybe a firewall rule could break a business function or an automated response could cause just as much disruption as the attack it was intended to repel.\n\nPlus, cyber defence is one of many priorities competing for money and resources inside an organisation.\n\nChismon wrote that “some board members may see little difference between a DoS attack taking down the organisation’s IT, or a poorly implemented action by the cyber defence team that does the same thing.”\n\n“Using AI automation for defence therefore quickly becomes a matter of organisational politics, and someone needs to be responsible for the action taken,” Chismon wrote.\n\nThat means defenders have to establish how much freedom an AI system can safely be given before allowing it to act.\n\n## Defeating autonomous attacks\n\nThe NCSC proposes a framework for estimating the risk of autonomous defensive actions based on five factors: potency, scope, criticality, rollout confidence and recoverability.\n\nAt the lowest end, an AI system might provide explainable advice to a human. At the highest end of the framework, automated systems could execute code or directly change systems, operate across an [enterprise](https://www.machine.news/tag/enterprise/), affect [critical systems](https://www.machine.news/tag/critical-systems/) or services, and make changes that could be difficult or impossible to reverse.\n\nFor now, Chismon identifies lower-potency tasks as an easier place to start, particularly where AI advises humans rather than directly affecting systems.\n\n“AI is good at summarising and so is already useful to defenders in finding details, or making sense of the huge volumes of reports or threat intelligence,” he wrote.\n\nBut Britain wants to go much further. The NCSC and the Department for Culture, Media and Sport are jointly working on Cyber Shield, which Chismon describes as a “national-scale agentic cyber defence ecosystem.”\n\n“Delivering that vision will require making autonomous defensive actions possible in a way they currently are not,” Chismon wrote.\n\n## The NCSC's little book of pain\n\nThe NCSC is also preparing an AI for Cyber Defence “problem book” setting out areas where it believes further research is needed.\n\nOne central question is whether organisations can prove in advance that apparently low-risk autonomous actions really are low risk.\n\nChismon asks whether AI could analyse traffic logs and establish conclusively which routes clients use, or reverse-engineer systems and software binaries to determine exactly which network calls they could make and which processes they might need to spawn.\n\n“Answering these questions will give us, and the organisations we protect, the confidence to take automated defensive actions,” Chismon wrote.\n\n## READ MORE: [The UK has no power to stop dangerous AI models being unleashed, Parliament warns](https://www.machine.news/the-uk-has-no-power-to-stop-dangerous-ai-models-being-unleashed-parliament-warns/)\n\nHe argues that solving these problems could also make it possible to automate system hardening, reducing attack surfaces and exposure as AI-enabled attacks become more capable.\n\nBut organisations cannot simply wait for autonomous cyber defence to arrive.\n\n“All these efforts will take time, effort, and research,” Chismon wrote.\n\n“Organisations cannot risk just waiting for agentic defence to roll in and protect them; they also need to be focussing on improving their security the traditional way.”", "url": "https://wpnews.pro/news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc", "canonical_source": "https://www.machine.news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc-warns/", "published_at": "2026-09-21 11:00:00+00:00", "updated_at": "2026-09-30 11:19:04.007367+00:00", "lang": "en", "topics": ["ai-safety", "ai-agents", "artificial-intelligence"], "entities": ["National Cyber Security Centre", "Dave Chismon", "Halvar Flake"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc", "markdown": "https://wpnews.pro/news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc.md", "text": "https://wpnews.pro/news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc.txt", "jsonld": "https://wpnews.pro/news/corporate-bureaucracy-could-give-ai-attackers-an-advantage-over-defenders-ncsc.jsonld"}}