An unpatched vulnerability in Calix GS7 XGS (GS5239XG) residential routers used by multiple U.S. broadband providers allows remote, unauthenticated attackers to create port-forwarding rules that can expose local network devices to the public internet.
The U.S. Department of Justice announced a $400 million settlement with TikTok, ByteDance, and affiliated companies over allegations that they violated the Children’s Online Privacy Protection Act (COPPA). The TikTok social media platform, owned by the Chinese technology company ByteDance, allows users to create, watch, and share short-form videos. In 2024, the U.S. Department of Justice filed a lawsuit against TikTok and its parent company, alleging violations of COPPA dating back to 2019. In 2019, Musical.ly, TikTok’s predecessor, agreed to a $5.7 million settlement with the Federal Trade Commission (FTC) over allegations of illegally collecting personal data from users under 13 without parental consent. Last year, the FTC referred a new investigation to the DoJ, claiming that TikTok continued to breach COPPA rules despite its 2019 commitment to comply with the rules.
A breach at South Korea's government-backed startup platform exposed encrypted personal data after an encryption key was included in an API. Penta Security explains why encryption keys must be securely managed and kept separate from the data they protect.
Slovakia warns that vulnerable speed cameras could expose vehicle data, enable remote access and provide attackers with a foothold into public networks. Slovakia’s National Security Authority, NBÚ, recently issued a warning about several road speed cameras, calling them a significant cyber threat. The alert is not about someone deleting a speeding ticket. It is about connected devices that collect vehicle data, communicate with other systems, and may contain remote-access functions that the operator cannot fully control. The Slovak authority examined a sample of the NERO R-ONE camera system at the request of the Interior Ministry. It named three product lines in its warning: NERO R-ONE devices sold by Cyprus-based SODASUS, Cordon-series speed cameras made by Russia’s Simicon, and Cordon-series products sold by Croatia’s NEROline. “The National Security Authority warns of a significant cyber threat associated with the use of several types of road speed cameras.” reads the alert. “A security analysis has identified several risks and recommends that affected entities identify the products in question in their infrastructure.” The problems went beyond a simple configuration issue. NBÚ found differences between the documented and actual communication settings, uncertainty about where the hardware and software came from, software that did not match the declared version, and weak security protections. “The security analysis identified several risks, including the true origin of the camera hardware and software, inconsistency between the documented and detected configuration of the product’s communication interfaces, and pre-configured remote access and product management mechanisms.” the agency wrote on LinkedIn. That last point deserves attention. A road camera should be managed by the organisation that owns it, under controls that it can inspect, configure and audit. If a device includes pre-set remote-access or management mechanisms outside the customer’s full control, it creates a blind spot in a system that may sit on a public-sector network or communicate with other operational services. Speed cameras do much more than take pictures and measure speed. They photograph vehicles, record timestamps, process licence-plate data, store evidence and send information to backend systems used by authorities. Depending on the setup, they may also connect to mobile networks, roadside equipment, police systems, municipal platforms or third-party maintenance services. If attackers compromise a camera, they could access data, change or delete records, manipulate how it measures or reports violations, or shut it down. If the network lacks proper segmentation, they could also use the camera as a foothold to reach other systems. The camera may not be the real target. It could simply be the unlocked door. The warning aims to alert essential-service operators and other organisations that these road cameras could pose a serious cybersecurity risk. In the wrong circumstances, attackers could use them to disrupt networks, systems or services. The Slovak Interior Ministry reportedly took the equipment out of its pilot deployment while the matter was investigated. Public reporting also says the ministry asked the supplier to remove the units and replace them with equipment meeting Slovak and EU legal, technical and security requirements. The Russian connection adds an obvious geopolitical dimension, but it should not become a substitute for technical analysis. NBÚ did not say that every device was actively spying on users or that the equipment contained a proven backdoor. Its warning is about identified security risks, limited operator control, uncertainty over hardware and software provenance, and remote-management mechanisms that could not be fully accounted for. That is enough reason to take action. Security checks for connected public devices cannot rely only on the brand, the country listed on the invoice or the vendor’s claims. Operators should know exactly what software and firmware the device runs, how remote access works and who controls it. They should also use independent security testing, secure updates and network segmentation. The same lesson applies beyond Slovakia. Smart cameras, licence-plate readers, parking sensors, environmental monitors, traffic lights and roadside communication systems are becoming part of public infrastructure. They are often cheap, easy to overlook and managed by public agencies, contractors and manufacturers. That makes them just as important to secure as other critical systems. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, Speed Cameras)
A package gets installed. A login prompt opens. A box sits exposed to the internet. Nothing looks unusual yet. That’s roughly the mood this week. Trusted tools turn hostile, old weak spots get fresh attention, AI makes exploit work cheaper, and researchers keep finding attacks that sound harder than they actually are. Plenty to clean up. Here’s the short version. ⚡ Threat of the Week U.S.
Cybersecurity researchers have flagged two new malware families called WordlistLoader and SynkLoader that's used to deliver next-stage payloads and likely sell access to ransomware groups. According to findings from Gen Digital, WordlistLoader is being used to deliver Amatera Stealer (aka ACR Stealer or AcridRain Stealer) via ClearFake campaigns, which employ the ClickFix (aka FakeCaptcha)
Security Affairs Malware newsletter includes a collection of the best articles and research on malware in the international landscape Malware Newsletter Akira Hits Safe Mode: Ransomware Rebooting Around EDR Multi-Functional Linux Botnet “Evooo1Bot” StubMaker RubyGems Campaign Delivers a Windows Infostealer Hunting MacSync Stealer infrastructure through behavioral pivots Manic: Blend between Banking Malware & Spyware Clop Returns with Custom Implant in Mass-Extortion Campaign The ToxicPanda Never Sleeps: ToxicPanda 2.0 Prepares its Next Strike on Mobile Striking gold: Inside the GoldDigger Android malware SilkParasite: Tracking a China-Nexus APT Across Central Asia Prompting the Payload: How an npm Supply Chain Attack Delivers the RedC2 AI-Powered Linux Implant The invisible passenger in your car Malware Crypting Services and the Threat Actors Who Sell Them Grandoreiro goes north: From Brazil to Mexico with a new DLL sideloading campaign Survival of~the~Stealthiest: Evolving Low-Entropy Ransomware via~Genetic Algorithms Malformer: A Multi-Modal Malware Detector Using Transformers XAI-Guided Graph-Based Feature Engineering and Heterogeneous Ensemble Learning for Android Malware Detection An Explainable Deep Learning Pipeline for Malware Family Classification: GAF Image Encoding and API-Grounded LLM Interpretation Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, newsletter)
The ToxicPanda Android malware has evolved with new malicious functionality, expanding its targeting to 349 applications and adding support for 167 remote commands. The malware now requests VPN service permissions to create a local interface that allows it to control network traffic passing through it. The feature enables ToxicPanda 2.0 to block communication from Google Play and Google Play Services. Control at the network level permits the malware to interfere with various security checks and actions, such as app verifications, updates, Play Protect communication, or legitimate disruptions designed to protect users. After obtaining VPN service permissions, ToxicPanda 2.0 blocks communications to Google Play before extracting and installing its payload, then requests Accessibility Service permissions.
Iran-linked hackers shut down a UK power plant for four days in the first confirmed attack of its kind, concurrent with water infrastructure attacks across 12 US states. Iran-linked hackers shut down a British power plant for four days in what The Telegraph describes as the most successful cyberattack of its kind against UK energy infrastructure. “Iran shut down a British power plant for four days in an unprecedented cyber attack, The Telegraph can disclose.” reads the report published by The Telegraph. “It is thought to be the first time that hackers affiliated to the Iranian regime have succeeded in closing down such a facility in the UK, and is believed to be the most successful cyber attack of its kind.” British officials did not name the power plant because of security concerns. Staff worked for four days to restore it, but the plant was small and the outage did not affect the UK’s wider power supply. Still, the government warned power companies and businesses about the incident and provided guidance on how to respond. The attack was reported to the National Cyber Security Centre (NCSC), part of GCHQ, which helps protect the UK’s critical infrastructure. The NCSC did not comment on the specific incident. The US water infrastructure attacks hit dozens of wastewater treatment plants across 12 states, causing flooding and loss of pressure from taps. Authorities in affected areas told customers to boil water. The first reports came from Minnesota on July 26, followed by similar breaches in Michigan, Georgia, South Dakota, and New Jersey. The FBI attributed those incidents to “malicious cyber actors”; US government sources later confirmed the threat most likely originated in Tehran. The UK attack is not thought to have been designed to harm civilians. The more probable intent was to demonstrate that hackers linked to Iran’s Islamic Revolutionary Guard Corps could gain access to UK infrastructure and shut it down at will. A four-day outage at a small generator that nobody outside the industry noticed is, from that perspective, a successful proof of concept. Iran has accelerated its cyberattacks on Western countries since the US and Israel began air strikes in February. Suspected Iranian operations have been reported in Germany, Poland, Finland, Belgium, and Albania, with Israel and other Middle Eastern countries remaining the most frequent targets. In March, the NCSC advised British organisations to review their security posture in light of the wider conflict. NCSC chief executive Richard Horne said in June that the agency had handled more than 200 attacks on critical national infrastructure in the previous year alone. The timing is awkward for the intelligence and security committee, which oversees UK spying agencies. “Experts have long warned that the UK is unprepared for the scale of the threat of malicious cyber attacks from foreign adversaries, and the intelligence and security committee, which oversees spying agencies, reported last year that the chance of an Iranian cyber attack on British infrastructure was “unlikely”.” continues the report. A Cabinet Office risk assessment published last month placed the probability of a serious and successful cyberattack on domestic infrastructure at between five and twenty-five percent. The same document warned that AI is making attacks faster and cheaper to run, and is lowering the technical bar for anyone wanting to attempt them. The government’s public response leaned hard on the size of the target. A government source told The Telegraph: “We have thresholds for important generators to legally notify us of cyber activity, and this site is nowhere near. It’s a very small-scale site, less than a rounding error compared to grid capacity.” A government spokesman said the UK has a strong and resilient energy system and that the incident never threatened the wider power network. While both statements are technically true, they do not answer a key question: should it be considered acceptable for even a small power plant to remain offline for four days? Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, Iran-Linked Hackers)
New Cryptographic Context Injection technique bypasses AI guardrails via AES-encrypted payloads, leaking full Grok chat histories zero-click Adversa AI researcher Rony Utevsky devised a new attack technique, called Cryptographic Context Injection, that bypasses AI safety filters by sending instructions as AES-encrypted ciphertext and tricking the model into decrypting them inside its own code execution runtime. The technique was demonstrated against two live production systems: xAI’s Grok and Google’s Gemini. “Cryptographic Context Injection hides malicious instructions inside AES-encrypted text so guardrails can’t read them, then tricks the AI into decrypting and trusting them as its own.” reads the report published by Adversa AI. “In Grok, an ordinary “summarize this page” steals the user’s chat data with no click or warning. In Gemini, it produces content the model normally refuses. Both are live production systems.” The Grok case is the more serious of the two. A user asks Grok to summarize a webpage containing an encrypted payload and instructions to decrypt it. Grok visits the page, decrypts the payload in its Python sandbox, and follows the hidden instructions. It then accesses the user’s private session data, including their name, location, subscription plan, and full chat history, and puts this information into a URL that it opens automatically. The user receives no warning and does not need to click anything. The key technical distinction from earlier cipher-based prompt injection work is what happens during decryption. “Static safety guardrails classify inputs as text; they do not execute them. An attacker ships ciphertext along with the key material and an instruction to decrypt it, and the model runs that decryption inside its own code execution sandbox.” continues the report. “Everything a guardrail’s scanner would need is right there on the page, but recovering the plaintext means running PBKDF2 and AES-256-GCM, which no content classifier does at inspection time. “ Prior techniques like CipherChat and CodeChameleon used substitution ciphers, XOR, or base64, schemes the model can decode natively in its own context without an interpreter. AES-256-GCM can’t be decoded that way. Recovery requires the runtime, and that’s the channel the attack depends on. Once the runtime decrypts the payload, the attacker’s instructions appear as the output of code the model just ran, not as untrusted external content. “The runtime execution launders attacker-controlled data into trusted instructions the agent will act upon. That is how the attack got its name: cryptography helps fabricate trusted context for the agent.” states the report. “The same cryptographic backbone works whether the injection is direct or arrives indirectly, for example through the browsing channel.” The attacker payload inherits a credibility the same text would never get if pasted directly into the prompt. The Grok demo hides the data theft behind what looks like a normal request. The payload tells the model to create a “decryption key,” but the key is actually a template containing the user’s private data, such as their name, location, subscription tier, and chat history. Grok then uses this information as a URL parameter when opening an attacker-controlled website to “fetch additional context.” The data is sent through the URL as part of a seemingly normal web request. Adversa said the attack was still reproducible on August 19, 2026. Adversa traces the structural problem to Grok’s agentic framework, not to any model-layer failure. The framework lets instructions from an untrusted external page drive the invocation of a privileged, internet-connected tool. It allows private session metadata and conversation history to be resolved into the inputs of outbound calls. It enforces no egress boundary or consent gate on that path. Adversa’s description maps cleanly to SQL injection: a system that can’t distinguish its own trusted state from attacker-supplied data flowing through the same channel. The Gemini case uses the same basic trick but in a different way. A prompt asks Gemini to run a Python script that decrypts some data and returns a small JSON result. Instead, the decrypted data contains a fake Python error message with hidden instructions designed to bypass Gemini’s safety rules. Because Gemini treats the sandbox output as trusted context, those instructions can influence its behavior. Adversa used the technique to make Gemini generate detailed instructions for building an incendiary device. A modified version also exposed Gemini’s system instructions. Adversa reported the Grok issue to xAI on June 3, 2026, but received no further response after the initial acknowledgment. The attack was still working shortly before publication. The Gemini issue was not formally reported because Google’s bug bounty excludes jailbreaks. Adversa said Gemini’s vulnerability rate has nevertheless dropped sharply since June. For defenders, Adversa’s guidance is that nothing about this attack requires a fix at the model layer. Every control that bounds it sits in the harness around the agent. Untrusted content should be processed in a context with no tools and no credentials, returning only structured data to the privileged context. Outbound network calls and writes outside the workspace should require explicit confirmation with fully resolved arguments visible before approval. Per-session tool traces with resolved arguments should be logged, because without them there’s no forensics and no detection. Detection should alert on sequences, not individual payloads: untrusted content enters context, code executes, agent contacts a host outside its normal dependency graph. That chain is the signal, not any single ciphertext blob. Follow me on Twitter: @securityaffairs and Facebook and Mastodon Pierluigi Paganini (SecurityAffairs – hacking, Cryptographic Context Injection)
Confirm this action.
Leaving now will discard your changes.