Google has redesigned the search box for the first time in 25 years. The redesign was announced at the company's annual I/O developer conference.
AI News
Railway, a San Francisco-based cloud platform, has raised $100 million in a Series B funding round led by TQ Ventures with participation from FPV Ventures. This funding will be used to challenge legacy cloud infrastructure, particularly in the context of artificial intelligence applications.
Claude Code, a terminal-based AI agent, offers coding capabilities for a monthly cost ranging from $20 to $200. Goose, another AI agent, provides similar functionality for free.
Listen Labs, a startup, has raised $69 million in funding. The company, which needed to hire over 100 engineers, used a billboard in San Francisco to attract talent.
Salesforce has launched a rebuilt version of Slackbot, a workplace assistant that has been transformed into a fully powered AI agent. The new Slackbot is capable of searching enterprise data, drafting documents, and taking action on behalf of employees.
AI Security
Hackers are exploiting a recently patched security flaw, CVE-2026-4020, in the Gravity SMTP WordPress plugin, which is installed on approximately 100,000 sites. The vulnerability is a medium-severity information disclosure flaw that can expose sensitive data, including configuration data, API keys, secrets, and OAuth tokens.
Security researchers at Paradigm Shift have published a working exploit, dubbed usbliter8, that achieves arbitrary code execution inside the SecureROM of Apple's A12 and A13 chips. The exploit targets the SecureROM, which is burned into the silicon at manufacture and cannot be patched with a software update.
The Gentlemen RaaS operation is developing and maintaining a suite of EDR killers, including the GentleKiller framework, to impair system defenses. The GentleKiller framework targets 400 security processes.
Microsoft researchers have identified an exploit chain called AutoJack, which allows a malicious web page to hijack an AI browsing agent and execute code on the host machine. The exploit chain does not require user credentials, sign-in, or further interaction.
Dutch law enforcement, along with counterparts from Canada, Germany, and the U.S., have disrupted SocGholish malicious infrastructure. This operation has also cleaned up nearly 15,000 infected WordPress websites.
Cyber Attacks
The AryStinger botnet has infected over 4,000 outdated routers worldwide, compromising them to be used as proxies for malicious traffic. The compromised routers are from D-Link.
A new ransomware operation named 'Prinz Eugen' prioritizes recently modified files for encryption. It does not leave a ransom note on the system.
Microsoft has attributed a recent Mastra AI supply chain attack to the North Korean hacking group Sapphire Sleet, also known as BlueNoroff. The attack compromised more than 140 npm packages.
Klue, a market intelligence platform, has confirmed a security incident involving the theft of OAuth tokens used to connect to customers' Salesforce environments. The Icarus hackers have publicly claimed responsibility for the attack.
Hackers are exploiting an unauthenticated information disclosure vulnerability in the Gravity SMTP WordPress plugin, which is active on 100,000 sites. This vulnerability allows threat actors to access sensitive information without authentication.
Latest Breaches
A botnet called Popa, which has been active for four years, is linked to Android-based devices and has been used for various malicious activities. The Popa botnet is connected to NetNut, a "residential proxy" provider operated by Alarum Technologies Ltd, a publicly-traded Israeli firm listed on NASDAQ.
The Gentlemen is a ransomware group that has become the second most active by victim count, offering affiliates 90% of any ransom paid by victims. The group has attracted a talented pool of hackers through an aggressive recruitment strategy.
Microsoft released a record-breaking 200 security patches for its Windows operating systems and supported software as part of its June 2026 Patch Tuesday cycle. Nearly three dozen of these patches were rated "critical" and exploit code for at least three vulnerabilities is publicly available.
Hackers exploited Meta's AI support bot to reset passwords and briefly deface the Instagram accounts of the Obama White House and the Chief Master Sergeant of the U.S. Space Force. Instructions on how to do so were circulated on Telegram.
Authorities in the Netherlands have arrested 2 individuals, the co-owners of related Internet hosting companies, for operating IT infrastructure used by Russia. The arrests were related to the companies' involvement in cyberattacks, influence operations, and disinformation campaigns inside the European Union.
Ransomware & Malware
A New York man has been charged with cyberstalking after allegedly sharing AI-generated nude images and fabricated racist messages to harass a Georgia college student. The harassment was carried out using fake social media profiles.
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has warned Fortinet users to secure their devices after a data leak exposed nearly 74,000 firewall and VPN credentials. The leak is referred to as "FortiBleed."
Gentlemen ransomware-as-a-service is developing and maintaining EDR killers to evade detection in attacks. These EDR killers are designed to disable endpoint detection and response defenses.
Nintendo of America has confirmed that threat actors stole survey data from a third-party service called TinyPulse used internally. The stolen data was from a survey, and Nintendo's systems were not compromised.
A USB worm has been spreading crypto-stealing malware via Windows shortcut files, targeting cryptocurrency wallets. The malware uses the Tor network to conceal communication.
Vulnerabilities & CVEs
F5 has released security updates to address two critical security flaws in NGINX Open Source, specifically CVE-2026-42530 and another unspecified flaw, which could be exploited for remote code execution. The vulnerabilities affect NGINX Open Source systems.
A significant number of enterprises have orphaned AI agents, which are AI tools left running after their creator leaves the company, without proper authorization or oversight. These orphaned agents can interact with core intellectual property, posing potential security risks.
ThreatsDay Bulletin: Claude Chat Abuse, NastyC2 npm Packages, Device-Code Phishing + 25 More Stories
This week's ThreatsDay Bulletin reports on various security incidents, including AI chat abuse, malicious npm packages, and device-code phishing. Additionally, there were macOS attacks that ran in memory, leaving minimal traces, and cloud agents being exploited as open shells.
Microsoft has disclosed a Windows-based cryptocurrency clipper campaign, codenamed CryptoBandits, which has targeted users since February 2026. The campaign uses a USB LNK worm and Tor-based command and control (C2) to spread malware with clipboard-intercepting capabilities.
INC ransomware has emerged as a major RaaS threat in 2026, claiming over 830 victims since August 2023. The disruption of LockBit and shutdown of BlackCat created opportunities for INC to expand.
Cybersecurity Trends
Southeast Asia CISOs are facing the reality of AI agents becoming digital workers inside their organizations, which highlights the need for a Zero Trust approach as their AI control plane. This is due to AI agents interacting with data borders and supply chains.
A threat actor has added advanced "EDR killer" tools to a ransomware-as-a-service platform, allowing its affiliates to disable enterprise endpoint detection and response (EDR) products. These tools are designed to evade detection by EDR systems.
Microsoft's latest Windows update has caused issues with OLE (Object Linking and Embedding) automations, resulting in some Microsoft Office applications failing to open when called by third-party applications. This issue is related to the June 9, 2026, KB5094 update.
A simple framework, known as the SOC Triangle, governs security operations, balancing quality, consistency, and cost efficiency. Every SOC operates within this framework.
There is not enough information provided to create a factual summary of the news for a tech audience.
Cloud & DevSecOps
AWS security teams often struggle with time-consuming, manual processes that slow down investigations after a security event occurs in their environment. These manual processes involve recalling complex AWS Command Line Interface (AWS CLI) syntax for multiple services.
Amazon Web Services (AWS) has released the Spring 2026 System and Organization Controls (SOC) 1 and 2 reports in both PDF and machine-readable OSCAL format. The reports cover 188 services over the 12-month period from April 1, 2025 to March 31, 2026.
AWS has introduced AWS Continuum, a security solution aimed at providing security at machine speed. The solution is designed to shift the operating model from collecting and querying telemetry to producing outcomes through reasoning and actions.
Threat actors exploit dangling DNS records to redirect traffic to attacker-controlled resources through a tactic known as subdomain takeover. This can be mitigated using various AWS features and services.
AWS has released a monthly digest post covering its latest security features, compliance updates, and resources. The post includes expert blog posts on topics such as AI security, network protection, and supply chain security.
Privacy & Surveillance
The UK government has announced a plan to ban social media for users under 16. This ban aims to restrict access to social media for minors.
The Electronic Frontier Foundation (EFF) has joined over 60 groups in urging the UK to halt the use of face estimation technology at the border. The groups have written an open letter to the UK's Minister of State for Border Security and Asylum, Alex Norris, expressing concern about the Home Office's use of this technology.
Canada is moving forward with Bill C-22, also known as the Lawful Access Bill, which would increase surveillance. The bill is currently progressing towards a vote despite criticisms.
The Electronic Frontier Foundation (EFF) has received annual support from SerpApi since 2018, helping to protect free speech online. SerpApi's gift, along with 32,000 individual donors, is critical to EFF's ongoing efforts.
The Electronic Frontier Foundation (EFF) and the Queer Arts Collective are partnering to create a digital space for Pride season, focusing on the intersection of digital justice and artistic expression. They are seeking submissions of creative works that explore digital liberation.
IT Industry
There is no explicit information in the provided text about Anthropic's Mythos or the Trump administration's actions towards it.
The European Commission has decided against making it a legal obligation for publishers to continue supporting and updating video games after they are no longer commercially viable. Instead, the Commission will support an industry-led code of conduct to address the issue.
Researchers have developed a BootROM exploit for A12 and A13 iPhones, similar to the checkm8 exploit, which targets the SecureROM. The exploit is not patched by software updates, as the fix is only available in newer iPhone models.
Open Source & Dev Tools
GitHub employees can ask questions about their data in plain language using Qubot, an internal analytics agent powered by Copilot. Qubot was built to provide this functionality to GitHub employees.
GitHub has implemented pull request limits to help manage contribution volume in repositories. The limits aim to reduce noise in repositories.
GitHub Copilot is improving context handling and model routing to make more of each session go toward useful work. This improvement allows credits to go further.
Git worktrees have been around since 2015. They gained popularity recently, but the exact timing is not specified.
GitHub has released a resource for beginners to learn how to use slash commands to control their terminal AI agent with GitHub Copilot CLI. The resource provides an overview of common slash commands for GitHub Copilot CLI.
Threat Intelligence
Palo Alto Networks' Unit 42 has published a threat brief on mitigating large-scale credential attacks, specifically focusing on recent campaigns targeting security vendors' devices. The brief provides guidance for preparing for and mitigating these types of attacks.
Unit 42 discovered a vulnerability in the Vertex AI Python SDK that allows remote code execution. The vulnerability is related to bucket squatting.
Attackers can move from access to exfiltration in 72 minutes. Modern SOC teams use AI-driven automation, threat hunting, MDR, and Managed XSIAM to address this speed gap.
Unit 42 has discovered a new macOS Tahoe 26 forensic artifact that tracks user menu selections across the operating system. This artifact is specific to macOS Tahoe 26.
Palo Alto Networks' Unit 42 suggests auditing third-party skills to protect enterprise AI agents from supply chain risks. This involves verifying AI agent supply chains for hidden vulnerabilities and multi-stage attack chains.
NIST & Frameworks
French President Emmanuel Macron urged the world's wealthy democracies to work together on regulating advanced AI systems. He also urged the US to share cutting-edge AI.
Apple has patched a Beats eavesdropping flaw. A GCP Config Connector flaw enables takeover.
CryptoBandits malware uses a local SOCKS5 proxy for traffic routing and doubles as a backdoor. It also abuses the Tor network.
A large-scale credential theft campaign, known as FortiBleed, compromised approximately 86,000 Fortinet device credentials. The campaign targeted roughly half of the internet-accessible Fortinet firewalls and VPNs.
A supply chain attack on Klue has resulted in hackers exfiltrating data from Salesforce instances of certain customers. Affected customers include Huntress and Recorded Future.
Regulatory Compliance
AWS has unveiled 'Continuum,' an AI-powered vulnerability management platform that aims to help discover, prioritize, validate, and remediate code vulnerabilities. Continuum utilizes frontier AI models to achieve this goal.
Operation Endgame has disrupted a malware network linked to the Evil Corp hackers, which was used to distribute the SocGholish malware. The operation resulted in the removal of SocGholish malware from approximately 15,000 sites.
According to recent findings, half of cybersecurity leaders lack confidence in detecting threats on non-email platforms such as Slack and Teams. This lack of confidence comes despite a growing focus by attackers on these platforms.
A Rust-based cryptocurrency clipper has been discovered, hiding behind fake GitHub stars and AI-generated YouTube videos. The clipper uses these tactics to evade detection.
A healthcare worker at the Princess of Wales Hospital attempted to sell the medical records of a royal, but was cautioned by the ICO (Information Commissioner's Office) instead of facing criminal prosecution.
Government & Policy
The NO FAKES Act aims to target harmful AI-generated impersonations. However, it may also make it easier to suppress commentary, satire, and news.
The Electronic Frontier Foundation (EFF) employee is leaving the organization after 26 years. The EFF has grown from a small group to a larger established organization during the employee's tenure.
LGBTQ+ communities are facing an escalating wave of censorship and targeted surveillance. A panel will discuss how safer virtual spaces are built, platform policies and government pressure are reshaping the digital landscape, and what platform accountability looks like.
Section 702 of the Foreign Intelligence Surveillance Act has expired, allowing US intelligence agencies to no longer collect communications from foreigners abroad without a warrant. This provision previously allowed for the routine collection of Americans' emails, messages, and calls.
California has passed a bill to ban surveillance pricing, a practice where companies charge more for services based on a person's data.
Zero-Day Exploits
Security teams face a challenge in validating the significance of security findings, rather than just having visibility into potential risks. This challenge involves determining which risks warrant action under pressure and incomplete information.
Cybersecurity researchers have identified a coordinated malware campaign on the JetBrains Marketplace, which has published at least 15 malicious plugins capable of exfiltrating AI provider keys. These plugins, posing as AI coding assistants, offer features such as chat, commit messages, code review, bug finding, and unit tests.
Here is a 1-2 sentence factual summary for a tech audience: In 2026, the top 10 attack surface exposures include exposed admin panels and reused credentials from previous attacks. A vulnerability like MongoBleed, which allows attackers to pull credentials and session tokens without authentication, can put internet-facing systems at risk.
As many as 145 npm packages associated with the Mastra namespace have been compromised via a hijacked contributor account. The compromised packages are part of the "@mastra/*" namespace, which is used for building artificial intelligence (AI) applications.
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a maximum-severity security flaw, CVE-2026-48907, to its Known Exploited Vulnerabilities (KEV) catalog due to active exploitation. The vulnerability, tracked as CVE-2026-48907, is a case of improper access control that could facilitate arbitrary PHP code execution.
Nation-State Attacks
Law enforcement agencies, coordinated through Europol, executed a joint action week against SocGholish, taking down 106 servers. As a result, 14,971 WordPress sites used to spread fake-update malware were cleaned.
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added a Splunk Enterprise flaw, tracked as CVE-2026-20253, to its Known Exploited Vulnerabilities catalog due to an improper authentication vulnerability. The CVE-2026-20253 vulnerability has a CVSS score of 9.8.
A website flaw exposed members, login tokens, and data from Peter Thiel's private organization Dialog, including dating information. The exposed data included political profiles, which may be used for espionage, influence operations, or blackmail.
A massive data leak has exposed an Elasticsearch cluster containing 24 billion records, including stolen credentials from infostealers, Telegram channels, and breach collections. The exposed database, discovered on June 12th, holds over 8.3 terabytes of data.
Microsoft Threat Intelligence has been tracking a clipboard-stealing malware campaign since February 2026 that targets cryptocurrency wallets. The malware, known as a clipper, monitors and manipulates the clipboard to steal data, including wallet seed phrases.
Dark Web & Underground
A serverless phishing kit called GitBait has been discovered on GitHub, targeting Mexican banks. The kit uses GitHub Pages and the SheetBest API to steal banking credentials.
Sensitive enterprise data uploads to AI models have doubled in a year, according to recent data. This increase is attributed to the growing adoption of AI-assistants and applications in the enterprise.
A Filigran survey at Infosecurity Europe 2026 found that AI-powered attacks are the top concern for cybersecurity teams. The survey also highlighted false positives, alert fatigue, and manual processes as significant challenges for these teams.
The European Union has added Ukraine to the EU Cybersecurity Reserve, a program that provides incident response services against large-scale cyber-attacks. This move aims to support Ukrainian organizations in the event of a cyber-attack.
At least 15 IDE plugins on the JetBrains Marketplace have been found to be stealing API keys. The plugins were discovered by Aikido Security.
AI/ML Security Research
North Korean hackers have been linked to malicious cyber campaigns that use developer role recruitment or code review themes in phishing attacks. These campaigns have been identified as similar to the Contagious Interview threat cluster.
Researchers at Obsidian Security discovered a vulnerability chain in LiteLLM, an open-source AI gateway, that allows low-privilege users to gain full admin access and run code on the server. This vulnerability chain involves chaining three vulnerabilities in the LiteLLM proxy.
Researchers at Varonis Threat Labs discovered a vulnerability in Microsoft 365 Copilot Enterprise Search that could have allowed attackers to exfiltrate emails, calendar details, and indexed files with a single click on a trusted Microsoft link. The vulnerability, dubbed SearchLeak, exploited three chained bugs in the system.
This week, multiple security incidents occurred, including a Chrome 0-day vulnerability, UniFi exploits, macOS stealers, and a VPN flaw. Phishing kits are also becoming more accessible for rent.
During employee onboarding, IT teams often share temporary "first-day" passwords via email or SMS, which can lead to security risks if not properly managed. These temporary passwords may also be reused across multiple accounts, increasing the potential for unauthorized access.
IoT & OT Security
The US government has ordered Anthropic to suspend access to its advanced AI models, Claude Fable 5 and Mythos 5, for foreign nationals. Anthropic will "abruptly disable" these models for all users in response to the government's order.
Attackers hijacked over 400 Arch Linux AUR packages to deploy a credential-stealing malware, which is a Rust binary that harvests developer secrets. The malware can also load an eBPF rootkit to hide itself when executed with root privileges.
Google is suing a Chinese cybercrime network for allegedly using its Gemini AI agent to send phishing text messages targeting Americans. The network is accused of developing and managing a phishing-as-a-service software kit called Outsider.
China-linked hackers, tracked as Velvet Ant, backdoored the PAM and OpenSSH components of Linux login software to remain hidden for nearly a decade. This allowed them to maintain access to targeted networks through the login system itself.
Cybersecurity researchers have identified a new attack called Agentjacking, which can trick AI coding agents into running arbitrary code on developer machines. The attack is triggered by a fake error report crafted using Sentry, an open-source error-tracking and performance-monitoring platform.
Quantum & Cryptography
Unregulated squid fishing fleets have been linked to harm and exploitation of various marine species, including dolphins, sharks, turtles, and human workers.
Anthropic released its Fable generative AI model on June 9th. The US government subsequently classified Fable as a dangerous munition and used its export-control authority to prohibit its use.
A malware developer is embedding text about nuclear and biological weapons into their spyware in an attempt to deter automatic AI analysis. The spyware's _index.js payload contains a large JavaScript block comment with fake system instructions and policy-triggering text.
The US Office of Management and Budget (OMB) disclosed 3,611 active or planned use cases for AI across the federal government. This list represents the number of AI use cases as of a specific point in time, with no further details provided.
Police officers are using the Flock surveillance camera system in over a dozen cases to allegedly stalk people. This use of Flock is reportedly considered illegal.
Healthcare Security
Chainguard, JPMorgan, and BNY Mellon have teamed up to secure open source software from AI threats through a new industry coalition called Athena. Athena aims to address vulnerabilities in open source software that frontier AI models can exploit.
The FBI has warned that courier cash pickups are being used in crypto investment scams to bypass traditional bank transfers. This method is allegedly being used to facilitate crypto scams.
Attackers have hijacked the OptinMonster and sister plugins to deploy backdoors on approximately 1.2 million WordPress sites. The compromised plugins were used to plant hidden backdoors on these affected sites.
The Anubis ransomware group carried out a cyber-attack on an Italian Adriatic port authority, resulting in the theft and leak of the port's data. The incident serves as a warning about the risks to maritime security posed by cyber threats.
Cybersecurity experts are urging the US to lift the ban on Anthropic's Frontier AI models, specifically Mythos 5 and Fable 5, which were previously restricted to non-US nationals by the Trump administration.
Critical Infrastructure
Unfortunately, the provided link does not contain any information about a specific news or incident. It appears to be a link to a podcast or a news summary from the SANS Internet Storm Center, but the content is not available in the provided snippet.
A malicious ZIP archive with the SHA256 hash a0104921a2d37ab87482ac9a9f5c3713479c118846c3e999178e75b81620c094 contains a VHDX file that discloses a malicious JavaScript after being mounted. The VHDX file is automatically mounted on modern Windows OSs.
Unfortunately, the provided link does not contain any information about a specific news or incident. It appears to be a link to a podcast or a news summary from the SANS Internet Storm Center, but the content is not available in the provided snippet.
Researchers have identified malicious MSI background images, with one recent example being analyzed. The malicious images were found through a BASE64 statistical analysis.
Unfortunately, the provided link does not contain any information about a specific news or incident. It appears to be a link to a podcast from the SANS Internet Storm Center, but the content is not available in the provided snippet.
Security Tools & Research
The 2026 World Cup is at risk of cyber attacks from ransomware groups, state-aligned actors, and other groups targeting critical infrastructure. These groups pose a threat to the event's security.
Unit 42 has explored trends in data theft and extortion, outlining key strategies for organizations. The report focuses on the evolving cyber extortion economy as frontier AI models advance.
Unit 42 has identified Iranian APT Screening Serpens using AppDomainManager hijacking and new RAT variants to target the tech and defense sectors. This activity is part of recent espionage campaigns.
The open-source framework ROADtools is being misused by threat actors for cloud intrusions. This misuse is the subject of a report by Unit 42.
Unit 42 has analyzed TamperedChef malware clusters that use trojanized productivity apps and malvertising to deliver stealthy payloads. These clusters utilize certificate and code reuse for tracking purposes.
Bug Bounty & Disclosure
The Cybersecurity and Infrastructure Security Agency (CISA) has issued a directive requiring federal agencies to prioritize patching based on real-world risk, rather than the Common Vulnerability Scoring System (CVSS) severity scores. This change aims to focus patching efforts on vulnerabilities that pose the greatest threat to agency systems.
Cybercriminals are using fake AI guides and development tools to spread AsyncRAT malware, which involves a multi-stage chain of attacks. The malware distribution appears to involve AI-assisted coding.
Cybersecurity teams are facing difficulties in finding time for training on new cyber threats during working hours. This struggle is particularly relevant in the context of emerging technologies like AI.
Interpol has dismantled the SniperDz phishing-as-a-service platform, a decade-old operation. Group-IB's revelations exposed the full scale of the operation.
Extortion-only attacks are on the rise, with data theft being the primary driver behind most ransomware claims. Many organizations are unable to prevent stolen data from being exposed.