Taggart

@mttaggart@infosec.exchange
7.4K Followers
865 Following
9.2K Posts

Displaced Philly boy. Threat hunter. Educator. #infosec, #programming #rust , #python  #haskell , and #javascript . #opensource advocate. General in the AI Resistance. Runs @thetaggartinstitute. Made https://wtfbins.wtf. Not your bro. All opinions my own. Dad. #fedi22 #searchable

Pronouns: He/him.

The Taggart Institutehttps://taggartinstitute.org
Bloghttps://taggart-tech.com
Codeberghttps://codeberg.org/mttaggart
YouTubehttps://youtube.com/taggarttech
GitHubhttps://github.com/mttaggart
Keyoxideaspe:keyoxide.org:G4ADJFWICZZZXGR4STZQVMBJNM

@mttaggart 0% in the case of organisations using shared services. It's also not simply a matter of reluctance to hire staff for many.

Time zone coverage, availability, cost and quality of staff as well as replaceability matter. It's a hell of a lot easier to sign a contract with an SLA than to manage all the variables that go far beyond staffing - spare parts, electricity etc are very different propositions in large parts of the world than in the US or Europe.

The internet being what it is, I see some readings of this as an absolutist stance against SaaS. Nope! There are lots of reasonable uses of services rather than assets. However, an allergy to owned tools, even when the control over the tool would be an object benefit to the org, is a convention that I believe has done more harm than good.
How much of the justification for SaaS really boils down to "We have no idea how to administer Linux systems and we can't/won't hire anyone who does?"

Project Gutenberg Needs Your Donation

Your contributions enable us to plan our budget, assist our volunteers, and, where urgently needed, hire people with specialized skills. We still have a lot of work to do and gaps to fill. Donations like yours give us a financial cushion to address some of those gaps. We are filled with gratitude for the efforts of those volunteers and contributors like you.

https://www.gutenberg.org/donate/

Project Gutenberg

Project Gutenberg is a library of free eBooks.

Project Gutenberg

You may be tempted to think of prompt injection attacks against language models as "social engineering." Resist this temptation.

Prompt injection is a mathematical attack against a non-deterministic system. Language may be the substrate, but the substance is numerical vectors. In other words, thinking of the attack as human language is a pointless limitation. The possibilities of what can go into the prompt to produce undesirable output are functionally infinite.

Poetry, context shifting, and other human-like attacks are only the beginning. What comes next is a weaponization of the linguistic form in ways that seem utterly alien to human readers. But to the models, it's all just elements in the matrix.

The time has come to decenter book reading and essay writing as the pinnacles of English language arts education.

The idea that we must replace reading and writing with "multimedia" anything is yet another symptom of education surrendering to industry.

I will run underground literature classes for any and all. I will teach you how to write like a human being instead of a corpo prompt engineer. Go and do likewise, if you can.

https://ncte.org/statement/media_education/

Media Education in English Language Arts

NCTE statement on media education for the knowledge, skills, and competencies needed for life in an increasingly digital and mediated world.

National Council of Teachers of English
markdownify-mcp Report | Notion

Summary

thorn-pheasant-6d8 on Notion

Here's a nice little project for someone with entry level web development skill! Our website was built before the HTML dialog element was a thing! We could use someone to update it and improve accessibility

https://github.com/elementary/website/issues/3884

You may be tempted to think of prompt injection attacks against language models as "social engineering." Resist this temptation.

Prompt injection is a mathematical attack against a non-deterministic system. Language may be the substrate, but the substance is numerical vectors. In other words, thinking of the attack as human language is a pointless limitation. The possibilities of what can go into the prompt to produce undesirable output are functionally infinite.

Poetry, context shifting, and other human-like attacks are only the beginning. What comes next is a weaponization of the linguistic form in ways that seem utterly alien to human readers. But to the models, it's all just elements in the matrix.

Not sure if this is public info, but the Penn Oracle data breach definitely involved SSNs and banking info. If you ever interacted with Penn financially, probably best to watch those accounts.