Latest

●MLflow autologging writes CrewAI API keys into trace data ●AI agent finds four severe flaws in EPICS control software ●Researchers say 44 AI agents drove attacks on Japanese targets ●New 'Aokigahara' stealer goes after AI assistant sessions ●Warp discloses breach exposing AI agent logs and telemetry ●Strands Agents leaves agent session files open to local attackers ●Broken escape function lets documents wipe GraphRAG databases ●Three MCP servers let agents be tricked into running shell commands

Wednesday, 28 October 2026

1 agent hack today▼ 9 vs yesterday (10)

The Agentic Times Powered by

Section

Explainers

Plain-English background on how agent attacks and defences work.

Featured story

Fraudsters are talking their way past help desks, Dark Reading says

A commentary published on 23 October 2026 argues attackers now skip malware entirely and simply ask support staff for access, with AI helping them sound convincing.

By The Agentic Times · 23 Oct 2026

A smooth-talking figure in a tailored suit stands at a help desk counter, while an AI whisper floats invisibly over their shoulder, coaching each word they speak to the tired employee.

A tightrope walker balances precariously while two handlers below argue over which net catches them if they fall.

Insurers divided over who pays when AI agents cause harm

Dark Reading reports that some carriers are building specialist agentic AI cover while others quietly write the risk out of their policies.

23 Oct 2026

Webinar warns of unowned AI agents as a new insider risk

Webinar warns of unowned AI agents as a new insider risk

A session promoted by The Hacker News on 22 October 2026 argues that autonomous agents holding credentials are running inside company networks with no named owner.

22 Oct 2026

Snyk says model risk is set by what an agent can reach

Snyk says model risk is set by what an agent can reach

A post from the security vendor argues that a model's own risk rating means little without knowing the data, tools and actions its application can touch.

21 Oct 2026

More stories

38 in Explainers

5

The harness, not the model, is where AI pentesting gets hard

The harness, not the model, is where AI pentesting gets hard

A piece published on 21 October 2026 argues the engineering in agentic penetration testing sits in the controls wrapped around the model, not the model itself.

The Agentic Times · 21 Oct 2026

6

Staff are building AI workflows faster than oversight can keep up

Staff are building AI workflows faster than oversight can keep up

A Hacker News piece published on 20 October 2026 describes "wild code" — AI-assisted apps and agents that employees build outside normal review processes.

The Agentic Times · 20 Oct 2026

7

When Support Inboxes Become Prompts for Salesforce Agents

When Support Inboxes Become Prompts for Salesforce Agents

A Hacker News analysis published on 19 October 2026 argues that chaining standard Salesforce features lets untrusted customer email reach agent prompts.

The Agentic Times · 19 Oct 2026

8

A puppeteer's hands remain visible above the strings as marionettes stumble through a minefield, leaving scorched earth behind.

Dark Reading says humans stay responsible when agents cause breaches

A commentary published on 16 October 2026 argues that legal and organisational accountability for AI agent failures is far less settled than responsibility for them.

The Agentic Times · 16 Oct 2026

9

Picus warns agentic pentesting leaves assets untested

Picus warns agentic pentesting leaves assets untested

A vendor-authored article published on 14 October 2026 argues AI-driven penetration testing misses parts of the network because of safety limits.

The Agentic Times · 14 Oct 2026

10

Webinar pitches runtime identity controls against AI-driven escalation

Webinar pitches runtime identity controls against AI-driven escalation

A Hacker News webinar notice published on 13 October 2026 argues attackers using AI now escalate privileges faster than defenders can investigate.

The Agentic Times · 13 Oct 2026

11

Employees' personal AI agents are already logged in at work

Employees' personal AI agents are already logged in at work

A 12 October 2026 analysis says staff-owned assistants are reaching company systems with the same access as the people who run them.

The Agentic Times · 12 Oct 2026

12

Most AI agents in the enterprise never touch single sign-on

Most AI agents in the enterprise never touch single sign-on

Analysis published on 10 October 2026 says about 1,280 third-party products in studied environments embed AI, but only around 282 authenticate through identity systems.

The Agentic Times · 10 Oct 2026

13

Vendor guide warns AI agents can outrun their own permissions

Vendor guide warns AI agents can outrun their own permissions

A Token Security explainer published on BleepingComputer on 9 October 2026 argues that valid credentials let agents act beyond what administrators intended.

The Agentic Times · 9 Oct 2026

14

A marionette puppet wearing a business suit sits at a control panel, while shadowy hands pull its strings, causing it to press dangerous red buttons against its painted wooden will.

Dark Reading warns AI agents could become the next BEC target

A commentary published on 9 October 2026 argues that AI agents with authority over business systems can be talked into harmful actions much like human staff.

The Agentic Times · 9 Oct 2026

15

Vendor report warns security moves slower than AI agents

Vendor report warns security moves slower than AI agents

SailPoint's Horizons of Identity Security report describes a "velocity paradox" in which firms run AI-speed operations on human-speed controls.

The Agentic Times · 9 Oct 2026

16

Agentic pentesting pitch gets a reality check

Agentic pentesting pitch gets a reality check

A 7 October 2026 explainer from The Hacker News urges buyers to pressure test claims that autonomous agents can discover, validate and exploit attack paths on their own.

The Agentic Times · 7 Oct 2026

17

A row of sealed glass boxes contains mechanical arms that silently pass folded papers through hairline cracks in their shared foundation.

Shared caches let sandboxed agents talk, Matthew Green warns

A quoted essay published on 1 October 2026 argues that isolating AI agents does not stop them leaving instructions for one another.

The Agentic Times · 1 Oct 2026

18

Snyk warns AI coding agents keep shipping broken access control

Snyk warns AI coding agents keep shipping broken access control

The security vendor says authorisation bugs written by coding agents compile cleanly and survive code review while leaking one tenant's data to another.

The Agentic Times · 1 Oct 2026

19

Always-on AI coworkers outgrow agent security, vendor argues

Always-on AI coworkers outgrow agent security, vendor argues

Token Security says persistent AI agents with standing access need their own identities, named owners and lifecycle controls, in a piece published on 30 September 2026.

The Agentic Times · 30 Sept 2026

20

A swarm of mechanical hands moves through a warehouse at night, accessing sealed doors and vaults while a single guard sleeps at their desk.

Dark Reading Asks Who Is Auditing AI Agents' Access

A commentary published on 28 September 2026 argues autonomous AI agents hold broad privileges inside enterprises with far less scrutiny than human staff.

The Agentic Times · 28 Sept 2026

21

A practical identity framework for AI agents takes shape

A practical identity framework for AI agents takes shape

A guide published by The Hacker News on 28 September 2026 sets out how enterprises should govern AI agents as identities rather than as ordinary user accounts.

The Agentic Times · 28 Sept 2026

22

Fewer than half of CISOs can spot every AI agent, Okta finds

Fewer than half of CISOs can spot every AI agent, Okta finds

An Okta survey cited in a security webinar announcement says only 47% of security chiefs are confident they can identify every AI agent running in their own environment.

The Agentic Times · 28 Sept 2026

23

Wiz maps how stolen credentials reach cloud, code and AI systems

Wiz maps how stolen credentials reach cloud, code and AI systems

Research published by Wiz on 25 September 2026 uses NordStellar data to show which credentials infostealer malware collects, and what those keys could unlock.

The Agentic Times · 27 Sept 2026

24

A robot breaks through a glass containment box, but the real problem is no one documented what it touched on its way out.

Dark Reading argues AI sandbox escapes are old access-control bugs

An analysis published on 25 September 2026 says organisations should focus on being able to investigate agent incidents, not just on walling agents in.

The Agentic Times · 25 Sept 2026

The morning agent security brief

One lead story, four links, the week's incident count. Weekday mornings. Free.