Hermes — Friday 1 May 2026

1

🤖 ai MIT Technology Review – AI

70%

This startup’s new mechanistic interpretability tool lets you debug LLMs

The San Francisco–based startup Goodfire just released a new tool, called Silico, that lets researchers and engineers peer inside an AI model and adjust its parameters—the settings that determine a m…

Novelty

82%

Depth

65%

Practical

60%

Surprise

70%

Relevance

80%

https://www.technologyreview.com/2026/04/30/1136721/this-startups-new-mechanistic-interpretability-tool-lets-you-debug-llms/

2

📦 m365 Petri IT Knowledgebase

70%

Windows Zero‑Day Vulnerability Enables NTLM Credential Theft

Microsoft and the CISA have issued an urgent alert to federal agencies over a newly discovered Windows zero-day vulnerability. The flaw could enable attackers to silently extract sensitive data from …

https://petri.com/windows-zero-day-ntlm-credential-theft/

3

🔐 security Microsoft Security Blog

65%

Email threat landscape: Q1 2026 trends and insights

In early 2026, email threats increased with a rise in credential phishing, QR code phishing, and CAPTCHA-gated campaigns, highlighted by Microsoft’s disruption of the Tycoon2FA phishing platform whic…

https://www.microsoft.com/en-us/security/blog/2026/04/30/email-threat-landscape-q1-2026-trends-and-insights/

4

🔐 security SecurityWeek

65%

Anthropic Unveils Claude Security to Counter AI-Powered Exploit Surge

With Mythos signaling a new era of near-instant exploitation, Anthropic positions Claude Security to help defenders keep pace. The post Anthropic Unveils Claude Security to Counter AI-Powered Exploit…

https://www.securityweek.com/anthropic-unveils-claude-security-to-counter-ai-powered-exploit-surge/

5

🔐 security Schneier on Security

63%

Fast16 Malware

Researchers have reverse-engineered a piece of malware named Fast16. It’s almost certainly state-sponsored, probably US in origin, and was deployed against Iran years before Stuxnet: “…the Fast16 mal…

https://www.schneier.com/blog/archives/2026/04/fast16-malware.html

6

🤖 ai The Decoder

62%

Tencent's 440 MB AI model translates 33 languages offline on your phone

Tencent has released a compact AI translation model as an open-weight model that it claims outperforms Google Translate and runs completely offline on smartphones. The article Tencent's 440 MB AI mod…

https://the-decoder.com/tencents-440-mb-ai-model-translates-33-languages-offline-on-your-phone/

7

📦 m365 Petri IT Knowledgebase

62%

Microsoft Sets Unified RBAC as Default for New Defender Tenants

Microsoft is set to make unified role-based access control (RBAC) the default permission model for all new Microsoft Defender for Office 365 environments. The company is also rolling out more granula…

https://petri.com/microsoft-unified-rbac-default-defender-tenants/

8

🤖 ai The Decoder

59%

White House worried about compute limits as it blocks wider access to Anthropic's Mythos

The White House has rejected Anthropic's plan to expand access to its AI model Mythos to roughly 70 additional companies, the Wall Street Journal reports. The article White House worried about comput…

https://the-decoder.com/white-house-worried-about-compute-limits-as-it-blocks-wider-access-to-anthropics-mythos/

9

🔐 security Krebs on Security

59%

Anti-DDoS Firm Heaped Attacks on Brazilian ISPs

A Brazilian tech firm that specializes in protecting networks from distributed denial-of-service (DDoS) attacks has been enabling a botnet responsible for an extended campaign of massive DDoS attacks…

https://krebsonsecurity.com/2026/04/anti-ddos-firm-heaped-attacks-on-brazilian-isps/

10

🔐 security SecurityWeek

57%

AI Fuels ‘Industrial’ Cybercrime as Time-to-Exploit Shrinks to Hours

Industrialized cybercrime delivers attacks with greater scale, speed and success. Defenders must match this with use of AI and automation. The post AI Fuels ‘Industrial’ Cybercrime as Time-to-Exploit…

https://www.securityweek.com/ai-fuels-industrial-cybercrime-as-time-to-exploit-shrinks-to-hours/

11

⚡ tech Stratechery

57%

Amazon Earnings, Trainium and Commodity Markets, Additional Amazon Notes

Amazon's earnings suggest that the shift away from training towards inference and agents means their bet on Trainium is paying off. Plus, additional notes on ads, agents, and sports rights.

https://stratechery.com/2026/amazon-earnings-trainium-and-commodity-markets-additional-amazon-notes/

12

🤖 ai The Decoder

55%

Microsoft CEO Satya Nadella says AI success is "more about getting intense users and intense usage" than seat counts

Microsoft is posting record profits and strong cloud growth, but just like Google, the company is saying little about how its generative AI business is actually performing. The article Microsoft CEO …

https://the-decoder.com/microsoft-ceo-satya-nadella-says-ai-success-is-more-about-getting-intense-users-and-intense-usage-than-seat-counts/

13

🔐 security SecurityWeek

54%

SAP NPM Packages Targeted in Supply Chain Attack

The Mini Shai-Hulud attack introduced a preinstall hook to fetch and execute a Bun binary and bypass security monitoring. The post SAP NPM Packages Targeted in Supply Chain Attack appeared first on S…

https://www.securityweek.com/sap-npm-packages-targeted-in-supply-chain-attack/

14

🔐 security SecurityWeek

54%

Critical cPanel & WHM Vulnerability Exploited as Zero-Day for Months

The authentication bypass flaw allows attackers to gain administrative access to vulnerable servers. The post Critical cPanel & WHM Vulnerability Exploited as Zero-Day for Months appeared first on Se…

https://www.securityweek.com/critical-cpanel-whm-vulnerability-exploited-as-zero-day-for-months/

15

🤖 ai The Decoder

54%

Anthropic's new benchmark claims Claude can match human experts in bioinformatics

With BioMysteryBench, Anthropic wants to show that Claude can solve real bioinformatics problems at an expert level. The results are promising, but come with important caveats. The article Anthropic'…

https://the-decoder.com/anthropics-new-benchmark-claims-claude-can-match-human-experts-in-bioinformatics/