|
Scooped by
JC Gaillard
onto Artificial Intelligence and Cybersecurity October 17, 2020 2:16 AM
|
Scoop.it!
Poisoning an AI system can be alarmingly easy. Attackers can manipulate the data sets used to train AI, making subtle changes to parameters or crafting scenarios that are carefully designed to avoid raising suspicion while gradually steering AI in the desired direction.
 Your new post is loading...
Scoop.it!
From
www
The flaw, which impacted Amazon, Anthropic, Google, Cursor and others, let the agent give the human false information on which to make decisions.
Scoop.it!
From
www
Left ungoverned, an organization ends up with dozens of agents hitting production systems directly, and no single place to see the traffic or stop it.
Diamondstone's curator insight,
July 7, 2:27 AM
The future of cybersecurity will depend on innovation, collaboration, and continuous vigilance. In an increasingly connected world, staying secure online is not an option; it is a responsibility shared by all.read more...https://pressearn.it.com/blog/?s=Technology+
Scoop.it!
As cybersecurity platforms embrace agentic AI, organizations must balance detection performance against the escalating costs of token consumption, deployment architecture, and AI credits.
Scoop.it!
From
medium
A lot of “robotic horse pulls a buggy” kind of operations — where everything is kept exactly the same as it was in 2003, but “AI SOC” tools are simply tacked on to perform some of the tasks.
Scoop.it!
AI pentesting finds high-risk flaws at 2.7x the rate of other systems, and two of every three serious AI findings stay open and unfixed.
Scoop.it!
Don't panic about new AI threats; hackers are still using the same old security gaps to get in, so use AI to help your team finally fix the basics.
Scoop.it!
From
www
Organizations today must determine whether an autonomous system should be trusted to execute a specific transaction at a specific moment under defined conditions.
Scoop.it!
For years, security teams had to choose between cheap, consistent or high-quality operations, but AI is finally letting us have all three at once.
Scoop.it!
Alert triage and basic investigation Tier 1 analyst tasks are disappearing, but other roles will boom. Here are the prime candidates — whether you’re reshaping your career or reskilling your SOC.
Scoop.it!
Researchers say a single poisoned document could slow AI agent systems by up to 148× and turn AI safety controls into an enterprise weak spot.
Scoop.it!
Security leaders don’t need to build a new model to secure AI agents, established standards already provide the blueprint.
Scoop.it!
From
www
As AI systems discover and exploit flaws at unprecedented speed, organizations are still deploying software they know contains security weaknesses.
Scoop.it!
AI agents aren't black boxes — they're models interacting with software tools. The risk lies in their overlap.
|
Scoop.it!
As agentic AI and autonomous agents trend in 2026, enterprises face a major new hurdle: standard IAM stacks aren't built to govern these non-human actors.
Diamondstone's curator insight,
July 11, 3:48 PM
As organizations increasingly adopt autonomous AI agents, managing non-human identities has become a critical cybersecurity priority. The "Agentic AI Identity: A 6-Stage Maturity Model for Non-Human Identities" outlines a structured framework that helps organizations securely manage AI agents throughout their lifecycle. The model progresses from basic identification and authentication to advanced governance, continuous monitoring, automated policy enforcement, and adaptive trust. By following these six stages, organizations can reduce security risks, improve compliance, and ensure AI systems operate responsibly. The framework highlights that AI agents should be treated as trusted digital identities with clearly defined permissions, accountability, and ongoing oversight, making it an essential guide for organizations embracing AI-driven automation.read more.https://pressearn.it.com/blog/?s=Technology+
Scoop.it!
AI-generated code risks are slowing deployment, as most companies write code with AI but hesitate to ship it to customers.
Scoop.it!
Teams are dealing with a truly dangerous problem — automation that works, but that no one understands.
Scoop.it!
From
thenewstack
As AI agents outnumber humans 144:1, legacy security is cracking. Discover how to secure your infrastructure with Agent IAM.
Scoop.it!
From
www
The industry just has to face that vulnerability management has been broken for years.
Scoop.it!
From
www
Today’s intruders rarely smash through the front entrance. They slip in through a neglected API, a stolen login, an overpowered service account or an AI workload nobody fully owns. After that, they move sideways fast, and they move far.
Scoop.it!
The Attack Surface Your Security Team Isn't Governing Yet
Scoop.it!
AI's unpredictable, dynamic nature is breaking traditional cybersecurity models, requiring leaders to shift focus from prevention to real-time runtime visibility.
Scoop.it!
From
www
Organizations need to break the infinite renewal cycle of AI learning from the flawed data of previous AI models.
Scoop.it!
From
www
Separate advisories suggest familiar scams are evolving alongside enterprise AI adoption, expanding the focus from software vulnerabilities to user behavior.
Scoop.it!
Anthropic’s Mythos is accelerating vulnerability discovery to machine speed, forcing the bug bounty industry and offensive security teams to adapt to a future where finding flaws is no longer the hard part.
Scoop.it!
From
www
There’s just more code than ever and developers can’t verify everything – we need to make all that code understandable.
|
