Artificial Intelligence | Breaking Cybersecurity News | The Hacker News

The SalesBleed findings in Salesforce Agentforce show that an AI agent's real authority is decided by whoever can put words in front of it. Enterprises need to govern agent actions by the trust level of their inputs, and treat an agent's voice in internal channels as a credential. Many enterprises accept text from anyone on the internet and store it inside a trusted system. A sales lead form is one of the oldest. Someone types a name, a company, and a note, and the record lands in the CRM. For years, the security concern with those fields was script injection. With AI agents reading the same records, the concern has changed shape. The risk is now malicious instructions. An agent's effective authority in any given moment is not the authority in its role definition. It is whatever its tools permit, directed by whoever authored the least-trusted text in its context.  A Web Form Becomes a Requester Researchers at Zenity Labs disclosed three vulnerabilities in Salesf...

For years the plan was simple: migrate off RSA and elliptic curves before a quantum computer arrives, and onto the post-quantum standards that will survive it. That plan now has a problem at both ends. AI is finding weaknesses in cryptography faster than people can, and the schemes you are migrating to are not exempt. In July, a post-quantum signature candidate that had passed two years of expert review was broken by an AI model in sixty hours and withdrawn the next day. Nothing deployed has fallen yet. But the question for a security leader is no longer which algorithm to move to. It is how fast you can move again when the next result lands. What cryptography actually rests on Almost everything you trust online depends on cryptography: the padlock in the browser, the signature on a software update, the authentication behind a payment. And cryptography depends on a handful of math problems that are easy in one direction and believed to be practically impossible in the other. Mul...

When a new threat emerges, the obvious response is to add another layer of protection. A new endpoint threat leads to a new detection tool, a new identity attack brings another monitoring system, while a new SaaS risk adds another source of alerts. Each tool fills a gap and adds another layer of visibility. The challenge starts when all those layers begin generating alerts at the same time.  In 2025, Kaseya monitored more than 27.6 billion SaaS events, of which almost 98.9% were low severity, but more than 278.9 million were medium- or critical-level alerts. With that much activity, finding the signals that need attention quickly becomes as much an operational challenge as a security one.  As the queue grows, analysts have to filter and prioritize. In the process, a signal that looks minor on its own can be overlooked even though it may become significant when combined with activity from another system. This is the alert fatigue paradox. Adding more tools can incre...

AI is changing software development by compressing work that once took days into hours. Code can be generated, tested, refactored, and documented faster, allowing development teams to deliver functionality at a pace that was previously difficult to achieve.  That acceleration creates an important challenge for security teams.  AI-powered software development can accelerate innovation, but it can also introduce hidden vulnerabilities when security implications are not clearly understood. Watch this webinar series on AI threat realities and proactive security readiness to gain firsthand knowledge of what makes this AI threat unique, coupled with remediation guidance for long-term resilience:  This series also explores how organizations can securely adopt and implement AI at business speed without losing control of cyber risk.  The risk is not simply that AI-generated code can contain vulnerabilities. Human-written code has always contained vulnerabilities....

For the past decade, cybersecurity has been built on assuming the breach. AI is now pushing the industry to focus on what has to happen before one. Attackers can create, adapt, and launch attacks faster than before. Frontier AI can speed up vulnerability discovery and shorten the time to exploitation. At the same time, employees in nearly every department, along with developers, are bringing in chatbots, coding assistants, local models, agents, and AI integrations across the business. For most organizations, especially those with lean security teams, this adds up to AI security overload: machine-speed attacks, more internal exposure, and more tools competing for limited attention. How Is Internal AI Adoption Expanding the Attack Surface? Internal AI adoption widens the attack surface through tools, endpoints, integrations, and data connections that security teams may not know exist. AI comes in through web and desktop applications, browser extensions, developer environments, ...

Insider risk used to be the line item nobody fought for in the security budget. That's changed. The annual cost of insider incidents at $19.5 million per organization - and that number keeps climbing. Money like that has pulled a lot of vendors into this space, and "insider risk management" now covers a wide range of products: some are barely more than activity logs with a dashboard, others are genuinely sophisticated. The label doesn't tell you much on its own. As National Insider Threat Awareness Month comes to a close, it's worth looking at how we evaluate the technical landscape. Here's what actually separates a platform worth buying from one that just looks good in a demo. Does it actually see the whole picture? People don't cause problems in one place. They email something from a personal account, then upload a file to a cloud drive, then plug in a USB stick two weeks later. If your platform only watches one of those channels, you're not ma...

It feels like every AI conversation starts the same way: "We need an AI strategy!" Leadership wants the productivity gains. Employees want the convenience. And IT is expected to make it all happen—securely, of course. AI can summarize meetings, translate conversations, draft emails, and surface information in seconds. It promises to eliminate the kind of repetitive work that quietly steals hours from every week. But beneath all the excitement lies a question that needs more attention. Before AI can help you work smarter, it needs access to your information. That could be a calendar invite, next quarter's product roadmap, or IP. Think of AI like hiring the world's smartest intern. It can research, summarize, and help organize your otherwise disorganized meeting notes. In other words, you have to hand it your notebook full of your confidential information. The real question is what happens to that notebook after you've handed it over. Shadow AI isn't a...

There are two common mistakes you might be making with ISO 27001: treating certification as the finish line, and treating certification as proof that your controls will keep working.   Neither of these assumptions works well as AI becomes a bigger part of business operations.  ISO 27001 helps set up the right processes, ownership, controls, and risk methods for an organization. But AI systems now connect to more data, applications, and business processes, and they often act with less human oversight.  Meanwhile, businesses aren't slowing down.  OneTrust's latest research shows that 86% of organizations had at least one AI-related incident last year, but only 27% slowed or paused their AI rollout.  For CISOs and risk leaders, just showing that controls exist isn't enough anymore. Teams now need to check if those controls still work, what risks remain as things change, and who is responsible if risks go beyond what the organization can accept.  Th...

As the 2026 election cycle hits full stride with midterms approaching, the threats that voters, campaigns, and candidates are coming face-to-face with have completely changed from even two years ago. In 2024, most synthetic media and generative AI attacks were still unpolished, easy to spot, and deployed in isolated experiments.  Fast forward to today, and AI tools have grown up fast. What used to be scrappy proof-of-concept attacks are now slick, automated operations running at scale. That shift changes everything about what election security actually means. It's no longer just about locking down voter rolls or hardening voting machines. The real fight has moved upstream, into the channels where public trust gets shaped in the first place. If you're a security leader, a campaign staffer, or someone running digital infrastructure, the old perimeter-defense playbook won't cut it anymore.  For campaign security teams, that means the job has expanded well beyond protec...

You cannot reliably predict what an AI agent is going to do. That is a feature, not a flaw.  When things go well, developers can save a lot of time and effort. When it goes wrong, it can be disastrous. For example, in April 2026, a Cursor agent working on a staging task for PocketOS encountered a credential mismatch, found an unrelated Railway API token with blanket GraphQL permissions , and used it to delete the production database and its volume-level backups in nine seconds.  Unlike deterministic scripts, which produce fixed outcomes given the same input, we use agents because they can take a loosely defined objective and determine the steps themselves. An agent can read context, choose tools, query systems, revise its plan, and take a path nobody explicitly programmed. Security teams can decide afterward whether those actions were acceptable, but fully enumerating them in advance is not only antithetical to using an agent but also practically impossible. On the ot...

AI agents have become incredibly useful across many industries by their ability to reach inboxes, documents, and financial information, then search, summarize, or act on what they find. But that access creates a data-in-use problem: conventional encryption protects information in storage and transit, but an agent generally needs it decrypted in memory while processing it, where sensitive material can be exposed to application code, logs and debugging systems, infrastructure operators, or a compromised host. Conan Yu's work offers one practical response to that problem. He is co-founder of Rena Labs , which develops infrastructure for confidential AI training and inference using trusted execution environments (TEEs), hardware-isolated environments designed to limit the surrounding host's access while code and data are processed. Over roughly two years, Yu has worked across hardware and cloud TEE deployments, privacy-preserving financial-data analysis, and private AI inference. He a...

Most security teams know that dwell time matters. The harder question is what to do about it. Dwell time is the period between an attacker gaining access and the security team containing the threat. During that window, a threat actor has time to learn the environment, steal credentials, move between systems, and reach sensitive data. For years, security teams have tried to reduce this window by adding more detection tools. The problem is that more alerts do not necessarily mean faster detection.  A recent industry incident response report puts the global median dwell time at 14 days, up from 11 the year before, quietly reversing a run of steady improvement that had held for close to a decade. The better way to think about it is as an operational problem. Two numbers matter most. Mean time to detect (MTTD) tells you how quickly the team recognizes a real threat, while mean time to respond (MTTR) tells you how quickly the team investigates and contains it. An AI native SIEM...

As organizations deploy AI coding agents across large monorepos and microservices environments, a fundamental problem emerges: the model may be capable of making the change, yet still lack the organizational context required to make the right change safely. A developer can ask an AI coding agent to deprecate an API field, update an authentication flow, or modify a service interface. The agent can inspect the code available on the developer's machine and search for references. What it may not know is that the field is consumed by four other services across separate repositories, that one of those services belongs to another team, or that the same field eventually carries sensitive data into a third party integration. This is not simply a context window problem. It is a code context problem: providing AI agents with accurate, current, organization wide evidence about how software actually behaves. One emerging approach is to generate that evidence directly from source code us...

AI coding agents are already inside engineering organizations. The problem security teams need to solve is not only that AI-generated code might be vulnerable. You already have ways to catch that: code review, CI, SAST, dependency scanning, and production monitoring. The real problem is that tools such as Claude Code, OpenAI Codex, Claude Cowork, and GitHub Copilot are becoming extensible agent runtimes. Skills, plugins, hooks, repository instructions, and MCP servers can influence what the agent reads, which tools it selects, what commands it runs, and where enterprise data is sent. Most AI governance programs stop at approving the application. Very few can tell you everything that has been installed inside it. That is the supply-chain gap. What changed: Third-party components are no longer participating only at build or deploy. They are participating in the agent's decision loop. From coding assistant to agent runtime The first generation of coding assistants mainly...

Many AI systems depend on input signals that teams cannot fully inspect or explain. These opaque sources reduce visibility into the data paths that influence model behavior. Engineers lose provenance records, limiting the diagnosis of abnormal outputs. This complicates the work of security teams that need clear records of what influenced a model at any point in time. Verifiable search data offers a stable alternative. It gives teams an input they can examine, store, and reproduce in controlled conditions. Engineers can compare model behavior against information that was publicly accessible at the time a result was produced, rather than depend on hidden internal signals. This article outlines why verifiable search data gives AI and security teams the clarity required to maintain operational control. Why Traceability Matters in AI Systems Traceability lets teams follow an input from its origin through each processing step. When every stage can be inspected, engineers can review...

When it comes to email security software, the cybersecurity industry has become very good at one thing: scoring the message that's already in the inbox. We've gone from blocklists to signature matching to behavioral ML, and each generation of email security innovation was a genuine improvement over the last.  But the losses keep climbing anyway. The median time it takes someone to click a phishing link is just 21 seconds after opening the email, and another 28 seconds to hand over credentials or payment data. That's under a minute start to finish. That number hasn't moved much in years despite everything we've thrown at the problem. It's time for a different approach. But first, we need to understand modern challenges. We're filtering messages, but attackers are running campaigns

Somewhere right now, malware running on a compromised machine is checking in with an AI model, asking it for a new version of itself. Google's Threat Intelligence Group caught this happening in late 2025. The malware, nicknamed PROMPTFLUX, does this every hour it runs, and each version comes back looking different from the last. By the time a security tool learns to recognize it, it has already changed shape again. This isn't a rare glitch or a lab experiment. It's a preview of how a growing share of attacks work today and why legacy SIEM platforms built to detect known patterns are starting to fall behind. Legacy security tools were built to detect, not to adapt A SIEM is a system that collects logs from every part of a company's network and looks for signs of an attack. For years, it worked like a security guard with a very long memory. The guard learns what a break-in looks like - a certain kind of file, a pattern of behavior, a code signature, and watches fo...

For a long time, the cybersecurity industry has framed social engineering as a psychological problem. We treat it as a battle of wits between a charismatic con artist and an unsuspecting employee. The prevailing wisdom says that if we just train our people to better identify scams or tear down malicious infrastructure slightly faster, we can stay ahead of scammers. But looking at the threat landscape from a purely threat-intelligence perspective tells us something entirely different: Modern social engineering is more than a psychological game. It's a highly optimized, industrialized deception economy. Attackers run campaigns like hyper-efficient businesses. They have Customer Acquisition Costs, operational budgets, and strict Return on Investment (ROI) targets. This is the part the industry doesn't like to sit with: If we want to truly break the social engineering attack chain , we have to stop focusing exclusively on building higher walls or executing reactive takedowns...

For years, our industry has treated "blocking" as the gold standard. If the email didn't land, if the malware didn't execute, if the alert fired in the SIEM, we called it a win. That mindset made sense in a world where most attacks came through a handful of familiar doors. But AI has changed the game. We're not dealing with hobbyists sending out clumsy phishing attacks anymore. Modern adversaries are running multi‑channel, AI‑assisted businesses at machine speed. And if all you're doing is blocking at the edge, you're not really defending. You're just delaying. Generative AI has made it trivial to spin up highly personalized, multi‑step social engineering campaigns that operate simultaneously across email, collaboration apps, mobile, social media, and paid media. The result is a social engineering attack chain : a sequence of stages designed to manufacture trust, erode judgment, and bypass brittle controls. You don't beat that by tuning another filter. You have to disrupt the at...

AI is changing the security landscape. More and more threat groups incorporate LLMs into their reconnaissance and exploitation workflows. The notion that some vulnerabilities are too complex to implement is now obsolete. Using LLMs, hackers can automatically find and exploit complex vulnerabilities. We have all heard of Claude Mythos and its ability to identify vulnerabilities in large codebases and exploit them automatically. But LLMs can do more than find vulnerabilities in code. ShinyHunters has scanned thousands of Salesforce Sites. They used a modified version of "AuraInspector". They possibly used an LLM to code their framework, mods, reconnaissance tools, and other aspects of their workflow. But the next step is to use AI to supercharge the attack process itself. We at Reco decided to explore what it would look like. Reco's security research team built an AI-powered agent capable of performing end-to-end security assessments of Salesforce Experience Cloud sit...