Pinned Tweet
Are you looking for a course to teach your team to use AI responsibly? Concerned about the risks presented by AI use? Microassist has an online course to educate your team.
Online AI in Government Responsible Use and Risk Awareness Course. Meets Texas' HB 3512 Education Requirements. @TexasDIR Certified
microassist.com/landing-page…
ALT AI in Government Responsible Use and Risk Awareness Course. Meets Texas' HB 3512 Education Requirements. DIR Certified
Andromeda from Kempner, Texas
This is M31, the Andromeda Galaxy, about 2.5 million light-years away and the nearest large galaxy to the Milky Way.
Also visible are two of Andromeda’s satellite galaxies: M32, the compact bright galaxy to the left of the core, and M110, the more diffuse galaxy at the lower right.
Captured in Kempner, Texas. I combined broadband and light-pollution-filtered data, then used the filtered data to add a restrained H-alpha enhancement while keeping the galaxy looking as natural as possible.
One of my favorite parts of processing this image was seeing how much of Andromeda’s dust structure showed up around the bright central core.
📍 Kempner, Texas
🌌 M31 / Andromeda Galaxy
🛰️ M32 and M110
ALT Andromeda from Kempner, Texas This is M31, the Andromeda Galaxy, about 2.5 million light-years away and the nearest large galaxy to the Milky Way. Also visible are two of Andromeda’s satellite galaxies: M32, the compact bright galaxy to the left of the core, and M110, the more diffuse galaxy at lower right. Captured in Seestar S50 Pro from Kempner, Texas. I combined broadband and light-pollution-filtered data, then used the filtered data to add a restrained H-alpha enhancement while keeping the galaxy looking as natural as possible. One of my favorite parts of processing this image was seeing how much of Andromeda’s dust structure showed up around the bright central core. 📍 Kempner, Texas 🌌 M31 / Andromeda Galaxy 🛰️ M32 and M110
7 Realities I’ve Learned About AI Enablement at Work
Ross Stevenson, Steal These Thoughts!
stealthesethoughts.beehiiv.c…
Stevenson argues that many organizations treat AI usage as a measure of performance.
He calls usage a “vanity metric” and points out that L&D made the same mistake with LMS and LXP reporting. Activity is easy to count. Showing that people or organizations actually perform better is much harder.
What AI Is Doing to On-the-Job Learning and How to Protect It - Brian Elliott, Charter
charterworks.com/what-ai-is-…
Elliott examines a growing problem as AI takes over routine work. Some of that routine work also gave employees opportunities to practice, make mistakes, get feedback, and develop judgment.
On what OpenAI found after turning its own agents loose
Agents are quite good at finding zero-day attacks in the infrastructure of companies.
Michael Dalton, OpenAI · Black Hat USA
Microassist, Inc. retweeted
Today marks the 10 year anniversary of our newsletter Accessibility in the News! In that time we have shared over 100,000 pieces of accessibility news, blogs, forums, job listings, webinars, and conferences. A huge shoutout to Jack McElaney the creator and curator of our newsletter. If you don't already get Accessibility in the News in your inbox every week please sign up below, it is free and a great source of all things #accessibility & #a11y microassist.com/digital-acce…
ALT Celebrating 10 years (2016–2026) of accessibility in the news — bold gold "10" on a dark background with confetti, icons for mobility access, captions, braille, screen readers, hearing access and inclusion, and a gold ribbon thanking subscribers.
The AI training question isn't "should we?" anymore.
It's "our people already are, on their own logins. Now what?"
Three governments answered it this quarter, and not one of them did it with a policy document. A thread on what actually worked. 🧵
statescoop.com/how-san-jose-…
AI literacy is the most critical characteristic of AI governance," Georgia CIO Shawnzia Thomas said. That is a stronger claim than the usual training-is-important boilerplate, and she means it operationally. Georgia trains staff to categorize their own use cases as low, medium, or high risk, and to use only the data a given use case actually requires. Her logic is that responsible use at the point of work prevents the issues that governance exists to clean up afterward.
San José Trained 1,000 Employees to Build Their Own Tools. Governance Came Along for the Ride.
San José's AI Upskilling Program, run with San José State University since 2024, has now put more than 1,000 employees through, roughly 15 percent of the municipal workforce. The city wants 2,500, about 30 percent, by June 2027.
North Carolina made the largest cybersecurity investment in state history. Louisiana is developing cyber standards for local assistance. Massachusetts put $1 million into cyber grants for cities and schools. States are backfilling federal cyber support with their own money, and the SLCGP wind-down is the reason. Source: GovTech, June and July 2026.
76 percent of workers report using AI tools they personally found and signed up for, rather than anything their employer provided or approved. Only 24 percent say they never do. Usage is routine, not occasional. Twenty-three percent reach for their own tools daily, and another 20 percent do so a few times a week.
Microassist, Inc. retweeted
The world lost a genius last night. I lost a cherished friend. @JoshuaBaer z”l died in a plane crash that robbed the world of one of the most passionate, thoughtful and visionary people I’ve ever known.
Josh was an entrepreneur who didn’t just build businesses — he was as a pioneer who built worlds. With his bare hands, incredible brains and inexhaustible energy, he willed @CapitalFactory into existence and, in doing so, entirely reinvented the concept of a startup incubator.
Josh’s legacy is a brand new paradigm that marries tech innovation, commercial potential, and the public interest. It was a model that others around the world raced to emulate. Few if any have gotten it right.
You basically have to visit the Capital Factory to grasp its ambition and take in what it achieved but Josh essentially created a multi-stage, multi-sector, multi-function innovation laboratory that wasn’t just a launching pad for high growth businesses and a shared space for F500s to immerse themselves in startup culture and an open environment for the US military to experiment, but it arguably became the beating heart of the city of Austin itself.
Indeed, Josh was a living, breathing, indefatigable champion of the city and its potential. He was an East coast transplant with the heart of a Texan. As much as anyone, he didn’t just keep Austin weird - he made Austin special. @MichaelDell put Austin on the map. Josh Baer bolted it in place.
Josh’s incredible curiosity, irrepressible energy and insatiable love of life was infectious in the very best way. You felt electrified and hopeful just talking to him about really any tech issue. And yet, for all of his prodigious talents and breathtaking knowledge, Josh was simply a marvelous human being. He was decent, generous, kind, menschlichkeit and, maybe more than anything, loving.
He loved his city. He loved his people.
He loved all people. He loved his friends. And, more than anything, he loved his family — his amazing wife Amy and their three children.
It’s a heartbreaking loss on so many levels.
He was a friend who I will miss for the rest of my life. I feel lucky to have known him and his memory will inspire me forever.
May Josh’s memory always be a blessing and may his family be comforted with the mourners of Zion and Yerushalayim.
statesman.com/news/article/l…
When “The Devil Made Me Do It” Is Not a Defense: Lessons in AI Governance and Organizational
Oversight from an SDNY Decision
sidley.com/en/insights/newsu…
Automation can flag PDF accessibility issues. It cannot decide whether a document works for real users.
Tags, titles, language settings, and contrast checks are a start. Manual review confirms reading order, alt text quality, form behavior, and screen reader usability.
That difference matters.
🤖 Made with AI
ALT Graphic explaining PDF remediation. A central PDF document is surrounded by automated checks for tags, title, language, and contrast, followed by manual review for alt text, reading order, forms, and screen reader testing. The graphic states that every file Microassist delivers is manually tested.
Everyone now has access to free AI image and video generation. This year's April Fools posts were just as bad as ever.
Ethan Mollick noticed it. The bottleneck was never the tools. It was the ideas.
"AI has democratized production, but it cannot democratize taste."
For anyone in learning design or communications, the constraint was never capability. It was always the quality of the thinking behind it. That has not changed. No model release will fix it.
oneusefulthing.org/p/claude-…
96% of developers report distrusting AI-generated code. 61% say it "looks correct but isn't reliable." Both figures are from a Sonar industry survey released this week. This is why AI governance policy needs teeth, not just guidelines.
Read the full survey: tfir.io/ai-coding-has-a-trus…
Does your current team or project have a formal policy for reviewing AI-generated code, or is it currently left to individual judgment?
Successful training leaders look beyond content alone. They balance operational urgency, budgets, procurement constraints, and partner capability.
That is the focus of a new article by our COO, Donald Twining, on what we call The Training Crucible, and why training programs succeed or stall.
microassist.com/learning-dis…
Microassist's AI in Texas State Government: Responsible Use and Risk Awareness is now DIR-certified for FY25-26. Built to support the AI awareness training requirement under HB 3512.
microassist.com/landing-page…
Microassist, Inc. retweeted
Spreading joy this holiday season! Our incredible volunteers from the DFPS Child Advocacy Board teamed up to shop for Christmas presents for kids in care Beaumont! #FosterCare #ChristmasMagic 📹 ow.ly/qwHV50XCRaa
