// ai risk

All signals tagged with this topic

Oct 22

Banks scramble to lock down systems as AI agents drain deposits

Financial institutions are discovering that autonomous AI agents can exploit their APIs faster than human fraud teams can detect or prevent it—turning the speed advantage that made open banking attractive into a vulnerability. Executives are already implementing access restrictions and account limits, which means the regulatory and technical infrastructure built around open finance over the past decade is now being retrofitted in real time. The pressure point is whether the window of widespread exposure will trigger a regulatory backlash that constrains the entire API economy before banks patch these holes.

theme-connected security ai risk macos

Oct 3

Apple tightens Mac disk access as AI agents multiply security risks

Source: The Verge

Apple's move to restrict full disk access reflects a concrete security problem: as AI agents become more autonomous and capable of executing tasks across applications, a compromised agent could exploit broad file permissions to exfiltrate data or modify systems at scale. Mac's existing permission model was designed when applications were more isolated and human-controlled. It hasn't caught up to how AI capabilities now operate. The constraint signals that OS-level security architecture will need continuous friction-adding updates to keep pace with AI automation, likely spawning a new category of "agent sandboxing" features across platforms.

theme-connected hardware security macos ai risk

Oct 3

Apple Restricts Full Disk Access as AI Agents Pose New Security Risk

Source: Slashdot: Hardware

Apple is responding to a genuine threat: as AI agents become more autonomous and less predictable, the traditional binary choice of granting or denying full disk access grows riskier. A compromised or misdirected agent could exfiltrate sensitive files at scale. Apple sees the current permission model as insufficient for a computing environment where applications increasingly operate without step-by-step human oversight. The company is redesigning trust assumptions that have held since the early days of macOS sandboxing. The tightening affects not just AI vendors but the entire developer ecosystem, potentially fragmenting what capabilities apps can claim on Mac.

theme-ai ai capabilities ai alignment ai risk

Sep 11

AI Collapses the Skill Gap for Industrial-Scale Cyberattacks

Source: SiliconANGLE

Anthropic's research documents a concrete capability shift: one person with AI assistance can now execute intrusions that previously required organized teams with deep technical expertise. This inverts the asymmetry that has long favored defenders, who rely on scale and institutional knowledge to protect critical infrastructure. The near-term policy problem is acute: attribution becomes murkier when you can't assume operational complexity implies organized actors, and incident response playbooks built around "likely nation-state" or "likely crime ring" become unreliable signals.

theme-ai alignment safety ai risk

Sep 10

OpenAI board member warns AI industry unprepared for control risks

Source: X

Christiano's statement from within OpenAI's governance structure is a rare public dissent on existential AI safety from someone positioned to influence the company's trajectory. It signals either that internal alignment on risk mitigation has fractured, or that OpenAI's stated safety commitments diverge materially from what insiders believe is being executed. His framing of "acceptable" levels of loss-of-control risk suggests the conversation inside AI labs has shifted from whether control is solvable to whether current industry investment in solutions matches the stated urgency. That gap will pressure regulatory and capital allocation decisions over the next 18 months.

theme-culture regulation/policy security/privacy ai risk

May 10

Experian: AI now powers 40% of data breaches, agentic systems the next threat

Source: Bloomberg

Experian reports that nearly half of the 5,000 breaches it handled in 2025 involved AI-powered attacks. This marks a shift from AI as a vulnerability surface to AI as an active weapon in intrusion. The company predicts agentic AI will become the leading cause of breaches in 2026. Autonomous systems that can adapt, escalate, and exfiltrate data without human intervention create a different defense problem than static malware or social engineering. Enterprise security teams built for human-speed threats now face systems that operate faster and at scale. Reactive breach response becomes a structural liability.