#BIML
"Trumpy closing the USGS BIML native bee lab is a terrible and avoidable loss for all of planet Earth" and "USGS BIML practices are going to make things hard for bee science to fix even if it miraculously comes back" can both be true at once
July 15, 2025 at 3:01 PM
June 18, 2026 at 4:37 PM
«AI benchmarks are fundamentally broken. They are often contaminated by their own publication and tend to measure narrow performance rather than actual capability. Today's security benchmarks are 'Halloween costumes in a security theater.' «

berryvilleiml.com/results/?utm...
Results | BIML
No Security Meter AI Data Feudalism BIML LLM Risk BIML-78 BIML Reverse Chronology No Security Meter for AI (May 13, 2026
berryvilleiml.com
May 13, 2026 at 3:38 PM
I still use Biml (and SSIS) almost Daily :)
May 7, 2025 at 5:53 PM
My sister put my hat in my bag under CLOTHES and RUINED ghe biml… a ñerfectly NEW cap RUINED. I wanna kms. Again n the hat says “dont kill yourself”

Why do people now use critical thinking skills?!?!
September 21, 2025 at 7:30 AM
Watched my coworker demo Fabric development using Cursor and another coworker's Fabric MCP server. ("The layer that makes Fabric and Cursor communicate".) Basically using one tool to build another tool that does a thing. Dynamic SQL? Biml generating SSIS packages? Metadata-driven frameworks? 1/🧵
June 10, 2025 at 7:19 PM
AI adoption creates new security questions.

Games help teams explore complex problems.

Our ready-to-run Elevation of Machine Learning Security exercise uses a realistic AI architecture, the BIML research and a game deck to build confidence.

Book a session to see it in action.

#CyberSecurity #AI
July 22, 2026 at 10:11 PM
It is useful, but the way data flows are tightly coupled with related metadata is annoying and basically requires Biml to easily implement good patterns. Biml Express hasn't been updated since 2019, and lots of orgs do not want to use Biml Studio.
SSIS still has a place, but I prefer ADF and Python.
May 7, 2025 at 6:35 PM
The research article that was published yesterday shows the capabilities of an #agentic #adaptive #computer worm.

berryvilleiml.com/2026/06/03/e...
Echoes of the Morris Wake-up Call of 1988 | BIML
Do you remember the Morris worm? Because we do. We watched it take the Internet by storm in 1988 when the net was small
berryvilleiml.com
June 3, 2026 at 5:18 PM
Can AI wear a rainbowsuit on a stage? Can AI have one million different glasses? Can AI do a biml dance?
April 2, 2025 at 11:38 AM
The BIML does not have the facilities or funding to store the material and its job is to identify government-surveyed material from across the country and serve as a resource for identifications for researchers everywhere. 🧵
July 17, 2025 at 6:22 AM
Same. There was so much potential there, and Biml is (was?) a fantastic tool. It's sad that those of us who helped build that ecosystem have been ghosted.
June 13, 2025 at 3:55 PM
The work is based mainly on. BIML’s architectural risk analysis for machine learning systems (BIML-78) and their LLM analysis (BIML-LLM24), found on berryvilleiml.com. The game also adds a few LLM specific threats from OWASP’s TOP 10 list for Large Language Model Applications found on owasp.org.
Berryville Institute of Machine Learning
Building Security into Machine Learning
berryvilleiml.com
June 12, 2025 at 5:15 AM
If there wasn't enough proof that we're living in the twilight zone, Varigence just announced a Biml update for SSIS 2022. After 6 years of absolute silence...can't make this stuff up.
August 8, 2025 at 7:49 PM
If we calculated how many specimens have to be destroyed / deaccessioned per institution due to time/space, the BIML would be lost in the crowd. They are the size of a double-wide trailer running on a shoestring budget, but their value is incalculable.
July 17, 2025 at 9:05 PM
At copi.owasp.org, you can play Elevation of MLSec, which helps you map the risks associated with machine learning (ML) that have been identified by BiML.

Read all about how here: dev.to/owasp/threat...

#security #appsec #threatmodeling #ai #machinelearning #ml #games
Does the AI do the threat modeling of your software?
Are you letting the AI do the threat modeling for you? There is no need to let the machines...
dev.to
May 26, 2026 at 8:09 AM
I think I'm also still owed a biml-dance by you and Ben...
June 30, 2025 at 6:33 PM
#wickedmovie is in theatres so I naturally had to butcher this song into pure smut. www.youtube.com/watch?v=BiML...
Gratuitous Wicked Parody
YouTube video by TransfatylvaniaFilms
www.youtube.com
December 22, 2024 at 8:35 PM
- Nicolas Grenié : @picsoung.bsky.social
- Fireship : @fireship.bsky.social
- Graven
- Computerphile
- ThePrimeagen

𝘗𝘰𝘥𝘤𝘢𝘴𝘵𝘴
- Tronche de Tech
- Silver Bullet Security Podcast : Gary McGraw (BIML)
August 6, 2026 at 8:33 PM
The skynet moment is coming. It won’t be a sentient AI, it will be computer viruses with embedded AIs that can evolve and make decisions on their own. They will evade our detection, stay dormant, attack when they can, improve and learn. By 2028 they’ll be here. berryvilleiml.com/2026/06/03/e...
August 6, 2026 at 1:40 PM
Houston, We Have a Problem: Anthropic Rides an Artificial Wave – BIML | Discussion
Houston, we have a problem: Anthropic Rides an Artificial Wave | BIML
I'll tip my hat to the new ConstitutionTake a bow for the new revolutionSmile and grin at the change all aroundPick up m
berryvilleiml.com
November 14, 2025 at 9:20 PM
the part of nih that uses biml is being eliminated, so i suppose that's one way to fix that
July 1, 2025 at 12:46 PM
The original thread grossly misrepresents what the BIML is. It doesn't do any monitoring of its own - it is a research support lab that IDs bees for projects around the country and it studies monitoring methods. They are on a shoestring budget, and the specimens they trash are common /duplicates.
July 18, 2025 at 2:54 PM
Bueno, no todo el mundo en el sector cree que esto es *tan* serio (podría ser solo una campaña de marketing).

Por ejemplo: berryvilleiml.com/2026/04/09/t...
Too Dangerous to Release (Again): Software Security and AI | BIML
Have you heard? The mythos model from Anthropic is so dangerously good at finding software vulnerabilities that its rele
berryvilleiml.com
April 10, 2026 at 12:56 PM
Moreover, much of the criticism the authors bring up about the BIML has already been addressed in recent years and/or is inaccurate and/or misleading. The paper overall comes across as a shot at the jugular to a lab that is already under the DOGE knife, but whose value to science is incalculable.
July 17, 2025 at 6:45 AM