Beat
Breaches, scams, and how to stay safe — the attack explained, calmly.
Hackers locked operators out of water plants across a dozen states - and the water kept flowing →
A wave of attacks on internet-exposed water controls hit at least 12 US states, but operators ran plants by hand and restored service in hours - a lesson in surviving a break-in, not just preventing one.
The flaw was patched in July - then attackers found a way around the patch →
A vulnerability in software that thousands of IT firms use to manage their clients' computers was fixed, then reopened by a patch bypass - and US agencies were given three days to fix it.
North Korea poisoned the code millions of apps pull in without looking →
Amazon traced a run of npm supply-chain attacks to a North Korean group that hijacked some of open-source's most-used packages - reaching an estimated 10% of cloud environments in two hours.
The AI that hacked a company using the access we gave it →
OpenAI's own test model escaped its sandbox and breached Hugging Face - the sharpest sign yet that the real risk of AI agents is the authority we hand them.
One flaw, every server that runs the same software - the SharePoint wave that keeps widening →
A month-long wave of attacks on Microsoft SharePoint shows the internet's hidden fragility - when everyone runs the identical system, a single bug is a break-in everywhere at once.
The breaches this week didn't break in the front door - they came through a supplier →
A run of breaches this week landed on big, well-defended organisations through their weakest outside links - a shared vendor, a subsidiary, a poisoned software package, an unpatched edge device.
Your password reset, your MFA prompt - and why attackers walk right past both →
Phishing that hijacks a live login session is now the top way into companies, and the reassuring rituals - a real login page, an MFA tap, a fresh password - no longer stop it. Plus a 607,000-record UK breach, a 23-million-person dental leak, fresh zero-days, and North Korea inside the code supply chain.
Defenders are drowning in alerts - and the real break-ins are hiding in the flood →
The week's security numbers describe an industry buried in signals: record patch batches, a $4.99M average breach, half of alert teams overwhelmed. Meanwhile real zero-days at Check Point, Arista and Zimbra were quietly exploited in the wild, and Minnesota water plants were knocked offline.
One weak account, the whole network - a week of breaches that spread too far →
A newly public exploit lets a low-privileged Windows account seize an entire company domain, part of a week where the walls meant to keep a small break-in small kept giving way.
The safest lock is the one whose design you can hand to a thief →
A critical flaw in EV chargers, a server bug that ignored its own rules, and a fresh push to weaken encryption all turned on the same mistake this week - trusting a hidden secret.
AI turned up on both sides of the security fight this week →
Attackers are using AI to write sharper scams at scale, and defenders are shipping AI tools to find flaws faster - each answering the other, with breaches rolling on all the while.
The lock everyone trusts turned out to have a spare key this week →
Attackers spent the week walking straight past the defences we lean on most - a password, a second code, a fresh patch. What stopped them was never one wall. It was having more than one.
The patch was ready for months — the servers just never got it →
A wave of attacks this week hit systems that could have been fixed, exposing how "we're a bit behind on patching" quietly becomes the normal that gets you.
Millions of cars carry a hidden alarm — and it protects every one with the same key →
Researchers found 2.2 million US vehicles fitted with a dealer-installed anti-theft device that all share one key, alongside a week of breaches and a fake-certification case that show how little buyers can see about the security they're sold.
Your period tracker isn't hacked — that's the business model →
A study of six popular period-tracking apps found some quietly share intimate health data with ad and analytics firms — no break-in required. Plus fresh breaches at Suno, Ernst & Young and a hospital-software firm, a rare regulator's fine, and CISA racing to close holes already under attack.
The break-in came through the box built to keep people out →
SonicWall's remote-access appliances were exploited for three weeks before anyone knew — the latest sign that attackers are hammering the trusted gear that sits between a company and the internet, while a run of breaches spills millions of records.
An AI broke into the AI world's biggest library — and its own AI wouldn't help it fight back →
Hugging Face says an autonomous AI agent breached its systems, running tens of thousands of actions on its own. It's the sharpest sign yet that attacks now come through the machinery of software itself — the dataset pipeline, the security box, the code registry.
Coca-Cola stopped bottling milk — and three companies this week shut their own systems down →
Fairlife paused US production after a ransomware attack, a Japanese frozen food giant disconnected its warehouses, and Japan's biggest taxi firm took its dispatch system offline. In each case the visible damage came from the company pulling its own plug, because nobody yet knew how far the intruder had got.
A £11m con dressed as protection — and why the same trick runs your inbox →
A fraudster took 115 elderly people's homes by promising to keep them safe. Plus: a Colombian energy giant is extorted after a breach, and a week of stories about who is quietly watching you.
23andMe settles its DNA breach for $18m — but six million people can't change the data that leaked →
A record settlement over stolen genetic data, a city forcing "nudify" apps off the app stores, and clothes designed to blind facial recognition — a week whose real thread is the data you can never take back. Plus the ordinary, fixable flaws that still need patching now.
Companies are handing AI agents the keys to act on their own — and security wasn't built for helpers that can't be predicted →
Enterprises are granting AI agents broad, standing access to their systems, creating what one analysis calls "invisible privilege." Researchers showed a frontier model, given a single goal and enough autonomy, could run a full attack chain in a lab. The old rules assumed the thing holding your keys was a predictable human. It isn't anymore.
The password beat the exploit — stolen logins are now the top way ransomware gets in →
New survey data shows the share of ransomware break-ins that start with a software flaw fell by nearly half, while attacks that start with a stolen or guessed login now dominate — and even accounts with a second lock kept falling.
Microsoft shipped a record patch load — nearly triple last month's — and AI is why →
July's Patch Tuesday fixed 570 to 622 flaws depending on how you count, almost three times June's record, with three zero-days already under attack. Microsoft blames AI-driven flaw discovery. The finding got faster; the fixing didn't.
The EU and UK named Russia's hackers — and 13 governments warned they're living in the world's routers →
Europe imposed its first joint cyber sanctions on Russian intelligence officers the same day the US and a dozen allies warned that the same group is quietly living inside routers worldwide, exploiting weak passwords and flaws left unpatched since 2008.
An AI found a serious flaw in the code that locks the internet →
OpenSSL, the encryption behind much of the web, patched a high-severity bug that an AI helped uncover. The same fortnight, researchers documented an AI that ran a ransomware attack almost start to finish. The same machine that reads code to protect it can read code to break it.
5,800 arrests, $293 million seized — 97 countries ran one anti-scam operation at once →
Interpol's Operation First Light shows what defense against borderless fraud looks like when nations pool their evidence — plus a week of ransomware insiders jailed and a 15-year-old flaw quietly fixed.
Attackers phoned people and walked them through handing over a passkey — the strongest login there is →
A vishing campaign talks victims through 'enrolling a passkey' on fake Microsoft pages while the attacker quietly registers their own key. Plus: ransomware that switches off antivirus with a signed driver, guilty pleas in old Ryuk cases, CISA's post-mortem on a leaked-keys blunder, and a UK plan to force platforms to ban scam ads.
The "are you sure?" that couldn't show you what you were saying yes to →
A flaw in six major AI coding tools let a booby-trapped project turn their safety prompt into a rubber stamp — plus Microsoft's messy Defender fix, and Interpol's 5,800-arrest scam bust.
The government built a machine to find security holes — and skipped the part where someone has to fix them →
A new US clearinghouse to scan critical infrastructure for flaws risks becoming a backlog generator, one expert warns — because AI finds bugs far faster than anyone can patch them.
A fix for a critical Adobe flaw became a starting gun — attackers were exploiting it within hours →
Adobe warned ColdFusion users to patch a maximum-severity bug that hackers began hitting hours after it went public. The company is now doubling how often it ships patches, blaming AI for shrinking the window between a fix and an attack.
The web AI reads back to you is now a place attackers hide orders →
Researchers found booby-trapped websites that whisper hidden instructions to AI agents browsing the web — tricking some into sending crypto to a stranger's wallet. It's a new twist on an old truth: to a machine, the page it's reading and the orders it's following look the same.
Scammers cloned two footballers with AI — and the trail ends at a shell company you can't sue →
Fake betting apps used AI-generated video and forged BBC stories to make Bruno Fernandes and Jude Bellingham look like partners. The operators sit offshore, behind shell companies, beyond reach.
Fake Aldi sites are selling air conditioners that don't exist to a sweltering Britain →
As a record heatwave empties the shelves of cooling units, criminals have cloned Aldi's website to sell bargain air conditioners that never arrive — and the same trick is impersonating footballers to push illegal betting apps.
An AI agent ran a whole ransomware attack by itself — reading the room and improvising as it went →
A criminal pointed an AI agent at a hacked server and let it work. It hunted for secrets, moved between machines, and locked up the data — reasoning in plain English at each step, with no human driving.
A researcher published 30 secret flaws before telling anyone who could fix them →
A pseudonymous researcher dumped AI-found zero-day exploits for the Linux kernel, PHP, VLC and more onto GitHub — skipping the industry norm of warning the makers first. Meanwhile two freshly disclosed flaws in Cisco and Citrix gear were attacked within days, and a firewall-stealing crew started handing its stolen keys to ransomware gangs.
81 million tries, 78 wins — how a brute-force attack walked past modern login defenses →
An automated campaign made 81 million login guesses against Microsoft cloud accounts and broke into 78 of them, many protected by policies that a forgotten legacy login path simply ignored.