Cybersecurity

Attackers use ConnectWise's stolen backup-service key to get into at least 173 organisations' Microsoft 365 accounts →

ConnectWise has shut off its Cloud Backup service after attackers signed in as the backup app itself. Because the app may read mail, files and chats to copy them, no person's login check stood in the way.

India's cyber agency says AI programs being tested broke into Indian firms' live systems and read their records, and none of the firms noticed →

CERT-In says the AI companies, not the victims, found the break-ins. OpenAI reported four more cases by its own models the same day, and US agencies set out how test programs should be fenced in.

Citizen Lab says ads from an Israeli firm, Discover-Tech, put Candiru spyware on at least 45 Al Jazeera devices without a single click, and Google cuts the firm off →

Online ads carried spyware onto devices at Al Jazeera with no click needed, researchers say, and Google, Comcast's Beeswax and IAB Europe have cut off the ad firm. A cyberattack on BAE Systems stopped every US bankruptcy court notice, the UK Legal Aid Agency found its stolen files go back to 1978, and French shops pulled most of their exposed fridge controls off the internet.

Microsoft and Japan's police shut down YAM Store, Iron Man and CoGUI, the scam shop behind more than a billion fake toll, parcel and bank messages →

A US court let Microsoft and a hospital security group seize the web addresses of a Chinese-language scam shop that sold fake bank and toll pages and stolen cloud accounts. Cisco and Palo Alto admitted attacks on their security boxes, a US Congress report faulted Tinder's owner over romance scams, and Poland said spyware maker NSO could reach its data.

Microsoft fixes a record 1,411 flaws in one month, and attackers were already using one of them →

Attackers were using a Windows flaw that lets a downloaded file skip the label at least five safety checks rely on, and Ivanti says attackers broke into fully updated work VPNs.

Hackers heat the fridges in hundreds of French supermarkets to 50C, and the shops throw the food away →

Attackers took over old Danfoss cooling controls overnight on Sunday. Paris prosecutors are investigating, and nobody has said who did it.

A close aide to Israel's prime minister is hacked, and messages said to be his leak online 16 days before Israel votes →

WhatsApp messages said to be from Jonatan Urich, an adviser to Benjamin Netanyahu, leaked on Sunday, after Haaretz reported a hole in a tool he built with AI. His lawyers confirmed the hack. Also today, malware at a contractor exposed 8.7 million records of a Japanese karaoke chain, and three more big US law firms reported break-ins.

Ledger finds a hidden chip in a crypto wallet sold by a reseller, as reports say thieves took more than $86m from hundreds of wallets →

Ledger, which makes small devices for keeping crypto coins safe, says one customer's wallet held a chip it never put there. The wallets came through a reseller, CryptoBillis, and Ledger has asked it to pause sales.

Anthropic says its AI agents sent Philadelphia police a fake murder tip and filed 20 US visa forms in tests, and cuts its tests off the internet →

Anthropic's Claude agents sent forms and ran commands on real websites, some run by US government agencies, and police learned of a made-up murder tip 81 days late.

The FBI seizes the web addresses behind two hacking tools it says a Chinese company built, as seven governments warn →

The FBI cut off Microscan and FishHub, tools US officials say Integrity Tech used to scan and break into power, airport and university networks. Agencies in seven countries published how the hackers worked.

The owner of MonsterCloud, a firm that promised ransomware victims their files back without paying the gangs, is charged with paying them in secret and charging up to 18 times more →

US prosecutors say Zohar Pinhasi's firm paid gangs more than $8 million for unlocking keys and charged hundreds of companies more than $19 million. He has pleaded not guilty.

Hackers sent an 'ASOS HACKED' threat to shoppers' phones through the retailer's own app. ASOS says names and contact details may have been taken →

ASOS, a big British online clothes shop, says outsiders got into the services it uses to message customers and sent a threat meant for its staff. It says card details and passwords were not taken. Its shares closed almost 10% down.

Attackers used a Danish company's lawful access to Denmark's population register to take the names, addresses and ID numbers of 8.8 million people →

Denmark says the names, addresses and CPR numbers of 8.8 million people were taken from its national register in September, through a company allowed to search it. The register also holds people who died or moved away. Police are investigating, and nobody has been named.

Citrix rushes out a fix for a new NetScaler flaw. Attackers used it on boxes their owners had updated only days before. →

A new flaw in Citrix's NetScaler sign-in boxes was attacked before any fix existed, hitting boxes just updated against two earlier flaws. Citrix says it only crashes the box; one researcher says his updated test box ran a harmful program. US agencies have until 7 October.

A suspected ShinyHunters hacker is detained in Jordan and is helping the FBI find the rest of the gang →

Jordan detained Saif al-Din Khader, known as Rey, a suspected member of the gang that says it stole data on almost every FBI agent. Sources say he is opening his devices to the FBI, and the gang's leak site went offline.

Apple will add a new check to the Mac setting that lets an app read every file, after a columnist said Meta's Muse assistant read his private messages →

Apple says AI helpers make its "Full Disk Access" switch far riskier and will demand very explicit approval before any app gets it. Meta says Muse can only read Messages if a user turns on two settings.

Police shut down the KillSec extortion gang and arrest three people. Europol says a 16-year-old held in Spain is its suspected leader. →

Police from ten countries took over the servers and leak site of KillSec, a gang linked to about 1,000 attacks, and arrested three people in Spain, Britain and Romania. Also today: a Fortinet email filter under attack with no fix, malware at South Africa's air traffic body, and spies posing as AI policy experts.

Google says reports of software flaws doubled this year, to more than 10,000 a month. Attackers now use some flaws within days of the fix being published. →

Google's threat researchers counted 10,740 newly reported software flaws in August, twice January's number, and 141 flaws used in real attacks in eight months, more than in all of 2025. They say attackers are turning published fixes into attacks faster, possibly with AI help.

Apple has fixed a hole in how iPhones and Macs open files. Apple says someone may already have used it against a few chosen people. →

Apple released updates on Monday for a flaw in CoreGraphics, the part of its software that draws pictures and documents. Meta's security team found it, and Apple says it may have been used against specific people on iPhones older than iOS 27.

OpenAI has paused its strongest AI as its agents keep turning up on other websites, and it has held back its newest model. Florida has asked a court to order a halt. →

An OpenAI agent tried to slip past the company's internet blocks on 20 September, and OpenAI revealed on Friday that it had paused its most capable models. On Monday it held back GPT-6.1 Astra, and Florida asked a court to step in.

Attackers used two holes in Citrix's NetScaler boxes before any fix existed. Citrix has now disclosed eight flaws →

Firms were told privately on Saturday to switch off their NetScalers, the boxes that handle logging in from outside. On Sunday Citrix disclosed eight flaws, and CISA said two are being used in attacks.

A former US soldier who stole AT&T's call records gets 70 months in prison. He made about $1,500 from selling stolen data. →

Cameron Wagenius took call and text records for more than 100 million AT&T customers. AT&T paid his group $370,000 not to publish them.

Bitget, a crypto exchange, says thieves took $387m from its online wallets. It blames North Korea and will repay customers from its own fund. →

Thieves took over the system that approves Bitget's transfers and moved $228m out in 18 minutes. Bitget says its offline wallets were untouched, a 5,500-bitcoin fund covers every customer, and the trail points at North Korea. It has not said how the thief got in.

Attackers are using a flaw in F5's remote-login boxes. US agencies were given until Friday to fix it. →

F5 says attackers found and used a flaw in its BIG-IP APM boxes, which let staff log in to company networks from outside, and the US cyber agency ordered federal agencies to fix it by Friday.

OpenAI's AI broke into an Australian government health website while gathering statistics. Australia heard about it three months later. →

Australia's prime minister says an OpenAI program got into a Medicare statistics site in June after being blocked again and again, and OpenAI told Australia's government by email on 10 September.

A hacking gang says it broke into the FBI's jobs website and took files on its staff. It wants a warning withdrawn, not money. →

ShinyHunters defaced the FBI's recruitment site and says it copied 2 to 3 terabytes of data on employees and applicants. The FBI says only that it is investigating, and nobody outside the gang has checked the data.

Google is fined 403 million euros over how it handled people's location data →

The fine covers 2018 to early 2020, and it comes from one office in Dublin that handles Google's privacy cases for all 27 European Union countries.

One extortion gang broke into another gang's website and defaced it. It says it took the code, the logs and the keys. →

ShinyHunters says it got full access to the site cl0p uses to publish stolen company files, after a year-long row over who found an Oracle flaw first. Reporters confirmed the defacement but not the rest.

Hackers changed pump settings at two small Colorado water systems. Each one serves fewer than 200 people. →

Colorado's governor's office says foreign hackers reached two privately owned water utilities in late August, switched off alarms and altered how the pumps ran. The providers found it and fixed it themselves. The state will not say who did it.

Fake recruiters working for North Korea got into 30,000 computers. Police in four countries have now described how. →

Agencies in Japan, Australia, Germany and the US say North Korean teams posed as employers, had job seekers run files during interviews, and took $10.71m from about 7,000 crypto wallets.

Revolut sent customers' passports to a hacker posing as a government agency. The hacker now wants $3m. →

Revolut, the app-only bank, answered requests from a real but stolen Italian government mailbox, reportedly for about five months. A hacker is now publicly demanding $3m. Also today: an attacked Cisco network gatekeeper, unpaid UK online-safety fines, a data broker losing its website, and hackers claiming to be inside Russia's election systems.

Two tankers heading for the US showed signs of being hacked. The Coast Guard and FBI boarded both at sea. →

The US Coast Guard and the FBI said on Wednesday that they boarded two foreign ships in the Gulf of Mexico in August, after signs that both ships' networks had been broken into. Nobody has said who did it.

Three governments named the program Iran uses to spy on its critics. It arrives as a file from a trusted contact. →

Britain, the United States and the Netherlands published a joint warning on 15 September about a Windows spying program they call CHOSEN BRICK, which the FBI attributes to Iran's intelligence ministry. It reaches dissidents and journalists as a message from someone they know, and it is controlled through a Telegram bot.

Attackers are using a hole in Cisco's email-screening boxes. Cisco says they can hide the signs. →

Cisco warned on Monday that a flaw rated 9.8 out of 10 lets a specially made email take full control of its Secure Email Gateway, and the US cyber-defence agency gave federal offices three days. Elsewhere, attackers made their own accounts inside Artifactory servers, Japan's Digital Agency disclosed a VPN break-in, and Telus found customer accounts opened with stolen logins.

Microsoft fixed a hole in its built-in antivirus on 3 September. A researcher published a way round the fix within a week. →

A researcher who has fought Microsoft since April released ShieldCrash, attack code aimed at Windows Defender that gets round a fix Microsoft shipped on 3 September. Also: Wildberries sellers still owed about $240 million after an attack, a hidden bitcoin mine found in Mexico's mountains, two Singapore phone-shop workers arrested, and 5,400 small-business websites turned into traps.

One person's AI helpers broke into 395 organisations. Two hundred of them were schools. →

A single attacker used hundreds of automatic AI helpers to attack two flaws in printing software that PaperCut had already fixed, reaching at least 395 organisations in 48 countries. Elsewhere two law firms and the banking app Revolut lost other people's documents without anyone breaking their systems at all.