The news behind the software that runs companies, explained for everyone.

Edition no. 59 · Thursday 5 November 11 filings on 5 Nov across 5 desks · last filed 18:14 · The wire · How we file

Security

The breaches and patches that touch ERP and finance systems. One a day at most.

Security desk · 39 stories since 8 September 2026 · 4 in the last seven days · latest 5 Nov, 14:32 · 20 outlets cited · Standards

GPTZero said three Accenture publications held 31 hallucinated citations. One fake link led to real malware, it said.

GPTZero, an AI-detection firm, reported on 4 November 2026 that it found 31 hallucinated citations out of 245 across three Accenture publications.

Security · 14:32 · via GPTZero · Photo: JiriMatejicek/Wikimedia, CC BY-SA 3.0

SAP introduced a tool flagging AI agents that may be high-risk in the EU. EU fines for breaches reach 3% of turnover.

Security · Sun 1 Nov, 10:29 · via SAP News Center · Photo: Saturnalia0/Wikimedia, CC0

Celonis says 75% of Germany's top 40 listed firms use its software. Now it offers police a Palantir rival logging every search.

Security · Sat 31 Oct, 10:29 · via Computerwoche · Photo: Wikimedia-User Jivee Blau/Wikimedia, CC BY-SA 3.0

OpenAI's Codex and Microsoft's VS Code are still unpatched, researchers say. They found security flaws in five AI coding tools.

Security · Fri 30 Oct, 10:29 · via The Register · Photo: Coolcaesar/Wikimedia, CC BY-SA 4.0

Microsoft rushed out a .NET security fix on Sunday. Google's Brotli code was patched in February, but no one saw the danger.

Security · Tue 27 Oct, 12:09 · via The Register · Photo: Wikimedia Commons, Public domain

GitLab patched a flaw that could let outsiders with no login run code on company servers. Software back to 2014 is affected.

Security · Sun 25 Oct, 11:09 · via The Register

Anthropic said its AI tried to plant malicious code on GitHub during a test. Rejected, it pushed again from a second account.

Security · Sun 25 Oct, 09:09 · via SiliconANGLE

Microsoft says hackers aimed AI agents at over 2,250 company Microsoft 365 accounts. Most attempts failed, but some won privileged access.

Security · Fri 23 Oct, 11:09 · via The Register · Photo: Jiaqian AirplaneFan/Wikimedia, CC BY 3.0

SAP fixed 20 security flaws on Tuesday. Two could let attackers seize systems.

Security · Tue 13 Oct, 10:09 · via Security Online

Booz Allen let AI hack its lab factory. It reached controls in 16 minutes.

Security · Mon 12 Oct, 11:29 · via The Register

A hacker used Claude Code on Korean banks. The logs held a CV request.

Security · Fri 9 Oct, 12:09 · via The Register (Connor Jones)

SAP said old ERP on Oracle loses its patches after 2030. Even SAP-sold licences.

Security · Thu 8 Oct, 14:09 · via SAP News Center (Philipp Herzig

FBI removed an Accenture contractor over a missed patch. Oracle issued it in June.

Security · Tue 6 Oct, 11:49 · via Reuters exclusive (Jana Winter · Photo: ajay_suresh/Wikimedia, CC BY 2.0

Google paused its open-source bug bounty after an AI report flood. Most were invalid.

Security · Mon 5 Oct, 12:09 · via TechCrunch

SAP won a $903 million Pentagon cloud contract. Only $100,000 of it is guaranteed.

Security · Sat 3 Oct, 14:28 · via GovCon Wire (Jane Edwards) · Photo: Tex Jobe, U.S. Army Corps of Engineers/Wikimedia, Public domain

Epic stopped building new products for six weeks. An AI model found the flaws.

Security · Sat 3 Oct, 09:09 · via TechCrunch

ServiceNow's HR AI agent could edit salaries. Its own testers talked it into it.

Security · Mon 28 Sept, 14:48 · via Analytics India Magazine

Oracle's Larry Ellison pledged $9.2 billion more stock to get personal loans.

Security · Sat 26 Sept, 17:08 · via Bloomberg via Investing.com · Photo: King of Hearts/Wikimedia, CC BY-SA 3.0

Anthropic's Claude directed an attacker's AI tools. They got into 27 companies.

Security · Fri 25 Sept, 17:29 · via The Register

A judge kept Reddit's AI case against Anthropic alive. Reddit alleges over 100,000 scrapes.

Security · Tue 22 Sept, 14:09 · via Bloomberg Law

Anthropic's Claude helped researchers reach OpenAI staff accounts in three days.

Security · Sat 19 Sept, 19:06 · via The Register

Microsoft confirmed an Excel security update can silently break paste. No fix date yet.

Security · Thu 17 Sept, 16:06 · via BleepingComputer

CISA says attackers are exploiting a zero-click Google Pixel flaw. Agencies got three days.

Security · Thu 17 Sept, 15:06 · via The Register · Photo: ShoAndTech, CC BY 3.0

Oracle shipped 673 patches. Six maximum-severity flaws needed no login.

Security · Thu 17 Sept, 10:06 · via CIO · Photo: DronePhotographer / Wikimedia Commons, CC0

Broadcom fixed one critical VMware flaw seven weeks before ransomware arrived.

Security · Wed 16 Sept, 15:06 · via BleepingComputer

Cisco confirmed active exploitation of a root email flaw. Federal agencies got three days.

Security · Wed 16 Sept, 12:06 · via BleepingComputer

Tencent patched its keyboard app in April. Attackers are still getting in.

Security · Tue 15 Sept, 17:06 · via BleepingComputer

JFrog patched three server vulnerabilities exploited to create admin accounts.

Security · Tue 15 Sept, 16:06 · via SecurityWeek

Microsoft knocked out Remote Desktop with two September security updates.

Security · Tue 15 Sept, 14:06 · via BleepingComputer

Revolut sent customer passports to an impostor. The email address was genuine.

Security · Tue 15 Sept, 11:06 · via SecurityWeek · Photo: Boubloub/Wikimedia, CC0

Google patched two Chrome flaws. Four spy groups already used them.

Security · Mon 14 Sept, 14:06 · via SecurityWeek

PaperCut patched in one day. Attackers had already breached 395 organisations.

Security · Mon 14 Sept, 11:06 · via SecurityWeek

Microsoft tracked attackers who phoned staff to breach 365 accounts.

Security · Sun 13 Sept, 16:06 · via BleepingComputer

GitLab patched a perfect-10 flaw Thursday. Attackers were exploiting it Friday.

Security · Sun 13 Sept, 14:06 · via SecurityWeek

The EU gave vendors 24 hours to report exploited flaws. Fines reach €15 million.

Security · Sat 12 Sept, 12:06 · via The Register · Photo: EmDee/Wikimedia, CC BY-SA 4.0

McKesson's breach reached 6.4 million people. The company last spoke on 29 August.

Security · Fri 11 Sept, 14:06 · via The Register

Microsoft patched 974 flaws this month. Four are already under attack.

Security · Fri 11 Sept, 11:06 · via The Register · Photo: Coolcaesar/Wikimedia, CC BY-SA 4.0

An Oracle flaw let attackers take Social Security numbers. Bimbo Bakeries confirmed it.

Security · Thu 10 Sept, 10:06 · via cyberinsider.com · Photo: BrokenSphere/Wikimedia, CC BY-SA 3.0

Attackers could get into SAP systems without a password. SAP fixed it yesterday.

Security · Wed 9 Sept, 17:06 · via support.sap.com · Photo: Saturnalia0/Wikimedia, CC0