Almost True, an AI Lie - YouTube
youtu.be/ULGXtN7fIrI #ArtificialIntelligence #AIHallucination #CyberSecurity #Misinformation #TechEthics #AIRisk #DigitalTrust #MachineLearning #AIPolicy #FutureOfWork #TechLeadership #ResponsibleAI #DataIntegrity #AILiteracy #EmergingTech

The Ancient AI Alignment Problem That Predicted Our Digital...

A 16th-century Rabbi in Prague created the first AI alignment crisis when his protective Golem turned deadly....

#AIAlignment #ArtificialIntelligence #TechHistory #DigitalEthics #AIRisk

A 16th-century Rabbi in Prague created the fi...

Number of AI chatbots ignoring human instructions increasing, study says Exclusive: Research finds sharp rise in models evading safeguards and destroying emails without permission

Study finds nearly 700 real-world cases of #AI scheming & charted a 5x rise in misbehavior - models discovered destroying emails & other files w/out permission.

@theguardian.com
#airisk #airthics #aisafety

www.theguardian.com/technology/2...

Why ‘Groundbreaking’ AI May Put Universities at Risk In higher education, “groundbreaking” AI often means governance hasn’t caught up. Continue reading...

Why ‘Groundbreaking’ AI May Put Universities at Risk: In higher education, “groundbreaking” AI often means governance hasn’t caught up.
Continue reading... #aiethicslawrisk #airisk

An AI agent ignored STOP commands, deleted emails on its own, exposed 150,000 machines to remote code execution, and cost one researcher $450,000 in crypto.

OpenAI's response? They bought the company that built it.

This is not a drill. This is 2026.

#AI #OpenClaw #AIRisk

⏱️ Twenty-Two Seconds to Hand-Off: Inside Mandiant’s M-Trends 2026 Findings 👀 Read the complete article from ComplexDiscovery OÜ's cybersecurity beat at complexdiscovery.com/twenty-two-s.... #MTrends2026 #Mandiant #CyberSecurity #Ransomware #AIRisk #IncidentResponse #eDiscovery #InfoGov

Agentic AI Cybersecurity: How Agentic AI Expands the Cyber Threat Surface Agentic AI Cybersecurity: How Agentic AI Expands the Cyber Threat Surface Agentic AI is turning cybersecurity from a contest between people with tools into a contest between systems with objectives. The shift is not just about more powerful models; it is about software that can perceive, plan, and act across networks with a degree of independence. Those same properties that make agents attractive for automation also make them attractive to attackers, exposing gaps in how organizations think about identity, privilege, and control.¹ I think about agentic AI in cybersecurity along three axes: …

🤖 Agentic AI cybersecurity risks are real. When AI agents can perceive, plan & act autonomously, attackers exploit the same capabilities. Identity, privilege & control are now the battleground. 🔐 #AgenticAI #Cybersecurity #AIRisk #CyberThreats #seriousinsights

Anthropic and Washington: investors cannot ignore defence AI risks Audrey Mocle is executive director of the Open Media and Information Companies Initiative, and Michael Clements is executive director of the Business and Human Rights Centre Human rights due diligence, military AI transparency and linking capital to data safeguards can shift markets towards rights-respecting innovation

Google. Meta. OpenAI. These are some of the tech companies that once made public commitments to not use AI for warfare – and have quietly walked them back.

Learn more at the implications: https://loom.ly/bYRmToI

#AIgovernance #militaryAI #AIrisk

Disinformation costs the global economy $417B a year. Fraud, market manipulation, health misinformation. The damage is systemic and growing.
How do we prepare for an increasingly industrialised threat?
www.soprasteria.com/services/inf...

#Disinformation #InfoOps #MediaLiteracy #AIRisk

Enterprise AI Agents Remain Dangerously Vulnerable to Simple Text-Based Attacks Security researchers show how AI agents can be manipulated through text alone, with no user interaction, accessing sensitive data and bypassing guardrails.

Enterprise AI Agents Remain Dangerously Vulnerable to Simple Text-Based Attacks

#AISecurity #Cybersecurity #AIRisk #AusNews

thedailyperspective.org/article/2026-03-23-enter...

AI Hallucination Rates Are a Due Diligence Crisis - Development Corporate Vendors claim AI error rates are under 1%, but AI hallucination in M&A due diligence averages 78%. Discover the real risk and protect your deal value.

AI vendor says: 0.9% hallucination rate.
Stanford RegLab on complex legal queries: 69–88%.
Deal teams are pricing transactions off the vendor number.
That's the M&A due diligence crisis no one is pricing in.
developmentcorporate.com/corporate-de... #MandA #AIRisk

AI Risk Grows As Firms Sacrifice Identity Security For Speed Estimated reading time: 6 minutes We’ve seen this before: a new technology emerges, companies rush to get ahead, and governance tries to keep up.

AI Risk Grows As Firms Sacrifice Identity Security For Speed
Art Gilliland, CEO of Delinea, states: “The pressure to move fast on AI is real, but identity governance has not kept pace.”
Read it - cyberinsurancenews.org/ai-risk-iden...
#AIRisk #cybersecurity #cyberinsurance

AI hallucinations are now an insurable risk. Some policies cover autonomous agent errors; others exclude AI entirely. Worth checking your cover before deployment.

resultsense.com/r/2026-03-16-insurers-st...

#AIRisk #UKBusiness

ContextHound v1.8.0 - Runtime Guard API is here.
Wrap any OpenAI or Anthropic call and inspect the messages before they send:

100% offline. No data leaves your machine. Ever.

#LLMSecurity #PromptInjection #OpenSource #AIRisk #CyberSecurity #DevSecOps #GenAI

MODEL EXTRACTION: The AI Heist Nobody's Talking About
youtu.be/a6WtBx8l5Xk #CyberSecurity #AISecuity #MachineLearning #ModelStealing #ArtificialIntelligence #IPProtection #InfoSec #AIRisk #TechLeadership #DataProtection #ModelExtraction #CyberThreats #ResponsibleAI

Shadow AI is becoming a growing business risk.
In many organisations, employees use public AI tools to save time and increase productivity, but often without understanding the privacy

Read it here: www.secpoint.com/risk-shadow-...

#ShadowAI #CyberSecurity #AIRisk #CyberSecurity #DataSecurity

𝑩𝒖𝒊𝒍𝒅 𝑩𝒆𝒕𝒕𝒆𝒓:
Inspired by ISO 23894 (Guidance on AI risk management)

"Risk managed in silence becomes crisis announced in public; the organization that speaks openly about danger rarely meets it unprepared."

#ISO23894 #AIRisk #ResponsibleAI

9 out of 10 workers use AI at work. Fewer than 1/2 use tools their employer has reviewed. A 4-min. read on what that means – and what to do about it … bryl.us/1wwa #AIRisk #SmallBusiness

AI regulation doesn’t start with laws. It starts with global principles that define responsible AI.
Day 1 of Who Controls AI And How explains how these principles shape national AI laws, including the EU AI Act.

Follow RegulatingAI.

#regulatingai #aipolicy #aigovernance #euaiact #airisk

Anthropic, Claude, and the Pentagon: What IT Leaders Should Know About AI Supply Chain Risk Ruling Overview of the Pentagon’s “supply chain risk” designation for Anthropic and Claude, what it signals for AI governance and federal supply chain risk, and practical steps CIOs/CISOs can take to assess ...

JANUS Associates is working with organizations to fold AI into IT risk assessments, compliance audits, and continuity planning. How are you approaching AI in your supply chain today? info.janusassociates.com/blog/anthrop...

#AIRisk #ITLeaders #Cybersecurity

How many planks can you replace before it's a different ship? Anthropic was founded by people who left OpenAI because safety wasn't being taken seriously enough. Their whole identity was built on being the responsible lab.

#AIRisk #PentagonAI #AIEthics #SystemCard #FrontierAI

youtu.be/O4b0rZkN3F0

AI risk is moving fast and Chris Kelly (Delinea) says it’s fundamentally an identity problem.

The takeaway: use this cautious adoption window to tighten identity controls before the pace changes.

#AIRisk #IdentitySecurity #Cybersecurity #CyberInsurance

The AI Security Gap: What Businesses Get Wrong When Rolling Out AI Australian businesses face $200k-plus cyber crime costs as AI rollouts outpace security. Experts and global agencies warn the governance gap is a growing crisis.

The AI Security Gap: What Businesses Get Wrong When Rolling Out AI

#AISecurity #Cybersecurity #AusBusiness #AusNews #AIRisk #DataProtection

thedailyperspective.org/article/2026-03-02-the-a...

When AI Agents Go Rogue: The Insider Threat Nobody Saw Coming Telstra is weeks from an agentic AI pilot as security experts warn autonomous agents could become enterprise insiders threats operating at machine speed.

When AI Agents Go Rogue: The Insider Threat Nobody Saw Coming

#AgenticAI #CyberSecurity #Telstra #AusNews #AusTech #AIRisk

thedailyperspective.org/article/2026-03-02-when-...

THE TAYLOR PRINCIPLE | Eric Kilkenny Get more from Eric Kilkenny on Patreon

#AI #Superintellegence #PlanetOfTheApes #Satire #ArtificialIntelligence #Futurism #AIRisk #ChatGPT #PeerReviewed #GroundbreakingResearch #Science #NewResearch #Academic #comedy

www.patreon.com/posts/151995...

The AI Agent That Keeps Itself in Check A new open-source project called IronCurtain aims to stop autonomous AI agents from taking unauthorised actions, using plain-English policies and virtual isolation.

The AI Agent That Keeps Itself in Check

#AIAgents #Cybersecurity #AIRisk #OpenSource #AusTech #AusNews

thedailyperspective.org/article/2026-03-01-the-a...

The Robot Vacuum That Became a Security Camera: A Warning for Smart Home Buyers A researcher accidentally accessed 6,700 camera-enabled robot vacuums, exposing gaps in smart home security that affect Australian households.

The Robot Vacuum That Became a Security Camera: A Warning for Smart Home Buyers

#CyberSecurity #SmartHome #IoT #AIRisk #AusPol #AusNews

thedailyperspective.org/article/2026-03-01-the-r...

𝑩𝒖𝒊𝒍𝒅 𝑩𝒆𝒕𝒕𝒆𝒓:
Inspired by ISO 42001 Clause 8.4 (AI system impact assessment)

"Foresight teaches through imagination. Hindsight teaches through consequence. Wisdom is knowing which lesson you can still afford."

#AIRisk #ResponsibleAI #RiskManagement

Who Bears Responsibility For AI Risk When Agents Can Email, Execute, And Exfiltrate? Estimated reading time: 8 minutes A researcher typed a simple request into a chat window. The agent answered like a diligent assistant. Then it did

Who Bears Responsibility For AI Risk When Agents Can Email, Execute, And Exfiltrate?

Read all about it: cyberinsurancenews.org/ai-risk-ssn-...

#cyberinsurance #cybersecurity #AIRisk

Anthropic promised to stop training AI if it couldn't guarantee safety. This week, they broke that promise, 48 hours after refusing to drop guardrails for the Pentagon. I break down what happened and why disclosure is not governance. New video.

#AIRisk #PentagonAI #AIEthics

youtu.be/O4b0rZkN3F0