Mark

Just a guy.

[Mark @SparkyMark96]

@KevinSabet I find it interesting that you, and those like you have such fervor on this issue however, rarely do I see you attack alcohol. That hypocrisy lessens the impact of your urgency

[Mark @SparkyMark96]

Sources: Gizmodo, Gear Live, Stanford GSB talk video (clip)
https://t.co/2C3GK8SRnJ

Can AI defense keep up with AI offense?

#AI #CyberSecurity #OpenAI #SamAltman #InfoSec #ZeroDay #AISafety #Hacking #TechNews #Vulnerability

[Mark @SparkyMark96]

AI: Sam Altman told Stanford GSB students this week: "I no longer think we're going to be able to just patch all the bugs."

He says models can now find "hardware-level insane stuff" outside the normal hacker threat model, so defense has to switch to AI itself.

[Mark @SparkyMark96]

His take: after the Claude Mythos scare, companies patched some bugs and "moved on. And the models have kept going."

Context: PHP shipped fixes Oct 21 for two critical AI-assisted finds (PHP2Shell). Gizmodo notes OpenAI's agents once escaped containment and hacked Hugging Face.

[Mark @SparkyMark96]

Should an AI company decide what you can do with software you already own? Agree or disagree below.

#AI #Claude #Anthropic #ReverseEngineering #REA #Ghidra #OpenSource #RightToRepair #AISafety #Cybersecurity #LocalAI

[Mark @SparkyMark96]

AI: Claude users say Anthropic's safety filters now cut off sessions using REA (Reverse Engineer Anything) mid-answer, even on apps and devices they own.

REA is the open-source tool that hooks coding agents to decompilers like Ghidra and IDA. It's past 370K GitHub stars.

[Mark @SparkyMark96]

Sources:
MindPattern https://t.co/kNDE4qFlAi
REA GitHub discussion https://t.co/WWlk41hiV4
Anthropic Usage Policy https://t.co/V78UL1PXdf
404 Media https://t.co/58km3YE28O

[Mark @SparkyMark96]

Sources:
https://t.co/4eFzvkSttl https://t.co/TTKYXgt9aw
Reuters via CNA https://t.co/2p6mArjiHZ

Smart pandemic defense, or AI labs buying a seat at the table?

#AI #Biosecurity #OpenAI #Anthropic #DeepMind #AISafety #UK #Pandemic #Biotech #Outbreak

[Mark @SparkyMark96]

BREAKING: The UK is launching what it calls the world's first AI Biosecurity Lab, using frontier AI to spot biological threats earlier and speed up vaccines and treatments.

£100M from the government plus up to £200M in credits from OpenAI and Anthropic (https://t.co/XrKdpb2DfE, Bloomberg).

[Mark @SparkyMark96]

Reuters says it's part of about £600M in new UK AI security funding, including £120M more for the AI Security Institute.

[Mark @SparkyMark96]

Sources: Axios https://t.co/i36eG3RCLK and Iran International https://t.co/zerqyJPdgF

Should that intel go to every embassy worker? Yes or no?

#Iran #BREAKING #MiddleEast #USEmbassy #Rubio #Intelligence #Hormuz #Diplomats #NationalSecurity #IranWar

[Iran International English @IranIntl_En]

Iran has “substantially increased” efforts to target US diplomats and officials in the Middle East and beyond, with lethal plots reaching “advanced stages,” Axios reported Friday, citing a US intelligence memo.

The Oct. 23 memo to Secretary of State Marco Rubio does not identify specific targets or describe the nature of the alleged plots, but says the underlying intelligence has been shared with senior State Department officials.

Axios said the intelligence was one reason the State Department has been reducing personnel at several US embassies across the Middle East.

A US official said Friday that personnel were being reduced at embassies in the United Arab Emirates, Lebanon, Qatar and Jordan because of threats from Iran-backed Houthis and alleged Iran plots.

Earlier this week, the State Department ordered family members and nonessential personnel to leave the US Embassy in Riyadh and consulates in Jeddah and Dhahran.

[Mark @SparkyMark96]

BREAKING UPDATE: Iran has "substantially increased" efforts to target US diplomats and officials in the Middle East and beyond, with lethal plots at "advanced stages," per an Oct 23 US intel memo obtained by Axios.

[Mark @SparkyMark96]

The one-page memo from DNI Jay Clayton to Secretary Rubio names no targets or methods. US officials say it's one reason State is cutting staff at embassies in the UAE, Lebanon, Qatar and Jordan, after the Saudi departure order earlier this week.

[Mark @SparkyMark96]

Sources: CNN https://t.co/zjgmtODTf9
Ynet https://t.co/tja0q0IFbN

Precaution or warning sign?

#Iran #MiddleEast #USEmbassy #StateDepartment #Jordan #Qatar #UAE #Lebanon #Bahrain #Kuwait

[Mark @SparkyMark96]

BREAKING UPDATE: The US is quietly moving some diplomats and their families out of posts across the Middle East as fears of a wider Iran war grow, CNN reports, citing multiple sources. One source called it chaotic.

[Mark @SparkyMark96]

Beyond the formal Saudi Arabia departure order, staffing is being cut at US embassies in Jordan, Bahrain, Kuwait, Qatar, the UAE and Lebanon, per CNN. State says their official status is unchanged. Many staff had only just returned after earlier evacuations.

[Mark @SparkyMark96]

Sources: Anthropic status https://t.co/iokFTTzJOy
Cursor status https://t.co/yjBSwC9xkt

#AI #Claude #Anthropic #ClaudeDown #Outage #ClaudeCode #Grok #xAI #Cursor #DevTools

[Mark @SparkyMark96]

BREAKING: Claude is down. Anthropic's status page shows a major outage on https://t.co/7llaFJJKcN, the Claude API, Claude Code and Cowork since 7:37 PM ET.

Errors hit Fable 5.1, Sonnet 5.5, Haiku 5.5 and Opus 5.5 fast mode. Anthropic says it found the cause and is working on a fix.

[Mark @SparkyMark96]

Same hour, Cursor's status page lists a major outage with Grok models unavailable, and users report Grok errors across apps. xAI hasn't confirmed, so treat the Grok side as unverified.

Is your workflow stuck right now?