Ian Linkletter's Avatar

Ian Linkletter

@linkletter.org.bsky.social

Emerging Technology & Open Education Librarian fighting for freedom of expression, access to information, and fair dealing rights.

1,591 Followers  |  890 Following  |  289 Posts  |  Joined: 02.05.2023  |  2.2412

Latest posts by linkletter.org on Bluesky

You can read the civil complaint and subscribe to updates at Courtlistener. www.courtlistener.com/docket/70989....

03.08.2025 03:06 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Thanks so much, I appreciate your support.

02.08.2025 04:58 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Yep! Those companies increased their prices long ago in response. Time to collect.

01.08.2025 20:59 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Hi Wes, thank you so much for mentioning me and including me in your notes! Could I please ask for you to remove the link to my slides and pad? I'm totally fine with everything else. Cheers!

01.08.2025 20:26 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Just put the finishing touches on my presentation for the Civics of Technology 2025 conference tomorrow morning: "Centering Student Voices in Resisting Surveillance". I can't believe I get to speak after @hypervisible.bsky.social! He has taught me so much. @civicsoftech.bsky.social #CivicsOfTech25

01.08.2025 06:34 β€” πŸ‘ 21    πŸ” 3    πŸ’¬ 1    πŸ“Œ 1

I am grateful to BCIT for including me on the AI Governance Committee! I'm going to focus on ethics and protecting students.

31.07.2025 02:22 β€” πŸ‘ 5    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Who has written the essay that talks about ai in terms of Blackness, literacy, and access to knowledge? Part of why I hate these systems so much is because I know the hard-fought battles just to use the library or learn to read. Now folks are telling us their machine has all the answers. πŸ€·πŸΏβ€β™‚οΈ

28.07.2025 19:07 β€” πŸ‘ 253    πŸ” 63    πŸ’¬ 18    πŸ“Œ 0
Preview
Setting the record straight: Scientists show that the algorithm that Proctorio used is incredibly biased towards people with a darker skin colour Do you remember when our Robin Pocornie filed a complaint with the Dutch Human Rights Institute because Proctorio, the spyware that she was forced to use as a proctor for doing her exams from home, co...

Setting the record straight: Scientists show that the algorithm that Proctorio used is incredibly biased towards people with a darker skin colour racismandtechnology.center/2025/07/29/s...

30.07.2025 03:18 β€” πŸ‘ 2    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0

Hat-tip to Hans de Zwart of the Racism and Technology Center: racismandtechnology.center/2025/07/29/s...

30.07.2025 03:13 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
A comparison of six individuals used to demonstrate face detection performance by OpenCV, with three dark-skinned individuals not detected ("N") and three lighter-skinned individuals detected ("Y"). Below each image are metrics: Estimated FTDR (Failture to Detect Rate), Skin Lightness (L*), Head Covering (Y/N), and Glasses (Y/N). The three undetected individuals have darker skin tones (L* values ~30–31), all wear glasses and head coverings, and have high FTDRs (~76–79%). The three detected individuals have lighter skin tones (L* values ~60–62), no glasses or head coverings, and very low FTDRs (~0.34–0.42%).

A comparison of six individuals used to demonstrate face detection performance by OpenCV, with three dark-skinned individuals not detected ("N") and three lighter-skinned individuals detected ("Y"). Below each image are metrics: Estimated FTDR (Failture to Detect Rate), Skin Lightness (L*), Head Covering (Y/N), and Glasses (Y/N). The three undetected individuals have darker skin tones (L* values ~30–31), all wear glasses and head coverings, and have high FTDRs (~76–79%). The three detected individuals have lighter skin tones (L* values ~60–62), no glasses or head coverings, and very low FTDRs (~0.34–0.42%).

OpenCV's "failure to detect rate" is highest for people with darker skin tones, people with head coverings, and people with glasses.

Racism, cultural discrimination, and ableism. Guess which remote proctoring tool contains all 3?

Research link: racismandtechnology.center/wp-content/u...

29.07.2025 23:36 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 1
Testing of AI systems is important for ensuring accuracy and re-
liability. In this study, we demonstrate how scenario testing with
demographically varied subjects, a form of prospective testing that
simulates real-world conditions, revealed significant performance
issues in biometric systems prior to broad deployment. Using gener-
alized linear modeling, we show that subjects’ measured skin light-
ness, along with other demographic factors, significantly impacted
the probability of failure to detect a face. Failure rates increased
from just 0.28% for subjects with the lightest skin in our sample
to 24.34% for subjects with the darkest, controlling for other fac-
tors. We show that skin lightness, rather than self-reported race,
best explained the differences in system performance. We trace
these issues to widely used, older methods in open-source packages
for face detection. Furthermore, this demographic differential is
not observed when testing open-source packages using a different,
more curated dataset. Our results highlight the need to evaluate
full multi-component, operationally deployed AI systems and the
role of scenario testing as a critical component of AI governance.
One way to mitigate the likelihood that poor-performing, older
open source methods are deployed in an operational system would
be to deprecate these functions in favor of higher-performing al-
ternatives. Prospective assessments of AI, in real-world use cases
with demographically varied subjects, should be used to identify
performance issues before these systems are operationalized.

Testing of AI systems is important for ensuring accuracy and re- liability. In this study, we demonstrate how scenario testing with demographically varied subjects, a form of prospective testing that simulates real-world conditions, revealed significant performance issues in biometric systems prior to broad deployment. Using gener- alized linear modeling, we show that subjects’ measured skin light- ness, along with other demographic factors, significantly impacted the probability of failure to detect a face. Failure rates increased from just 0.28% for subjects with the lightest skin in our sample to 24.34% for subjects with the darkest, controlling for other fac- tors. We show that skin lightness, rather than self-reported race, best explained the differences in system performance. We trace these issues to widely used, older methods in open-source packages for face detection. Furthermore, this demographic differential is not observed when testing open-source packages using a different, more curated dataset. Our results highlight the need to evaluate full multi-component, operationally deployed AI systems and the role of scenario testing as a critical component of AI governance. One way to mitigate the likelihood that poor-performing, older open source methods are deployed in an operational system would be to deprecate these functions in favor of higher-performing al- ternatives. Prospective assessments of AI, in real-world use cases with demographically varied subjects, should be used to identify performance issues before these systems are operationalized.

New research confirms previous findings that OpenCV, the open source facial detection software used by Proctorio, is racially biased.

Performance Differentials in Deployed Biometric Systems Caused by Open-Source Face Detectors: racismandtechnology.center/wp-content/u...

29.07.2025 22:40 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
"Responding to unacceptable AI use. AI detection tools including the one available in Turnitin have been found to be unreliable in university contexts, often producing false positives or negatives or easily being 
subverted. Unlike traditional plagiarism checkers, AI detection tools rely on statistical methods to estimate the likelihood that a piece of text was generated by AI. This can lead to wrongful accusations of academic misconduct, damaging trust between lecturers and students, and creating an adversarial learning environment. At UCT, the Senate Teaching and Learning committee has decided to discontinue the use of the AI detection score in Turnitin effective 1 October 2025."

"Responding to unacceptable AI use. AI detection tools including the one available in Turnitin have been found to be unreliable in university contexts, often producing false positives or negatives or easily being subverted. Unlike traditional plagiarism checkers, AI detection tools rely on statistical methods to estimate the likelihood that a piece of text was generated by AI. This can lead to wrongful accusations of academic misconduct, damaging trust between lecturers and students, and creating an adversarial learning environment. At UCT, the Senate Teaching and Learning committee has decided to discontinue the use of the AI detection score in Turnitin effective 1 October 2025."

University of Cape Town is discontinuing use of Turnitin's AI detector.

"AI detection tools including the one available in Turnitin have been found to be unreliable in university contexts, often producing false positives or negatives or easily being subverted."

www.news.uct.ac.za/article/-202...

24.07.2025 22:15 β€” πŸ‘ 5    πŸ” 4    πŸ’¬ 0    πŸ“Œ 0
Preview
Canvas is tracking your data. What is UBC doing with it? Proponents think it could revolutionize teaching. But some worry it’s a slippery slope towards an invasion of privacy.

From 2019: β€œShort filed a FOIA request, UBC refused, he had to appeal to the OIPC to get his own data. When the file arrived, β€˜It was something like 400,301,000 individual data points about me.’ Well beyond test scores & basic analytics: how he used his mouse, how & when he interacted with the LMS.”

24.07.2025 01:24 β€” πŸ‘ 64    πŸ” 33    πŸ’¬ 3    πŸ“Œ 4
"Instructure" with a red period at the end

"Instructure" with a red period at the end

Bluesky doesn't like SVG files. This is the new Instructure logo.

23.07.2025 23:02 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Canvas by Instructure logo featuring a circular red icon made of semi-circles and dots.

Canvas by Instructure logo featuring a circular red icon made of semi-circles and dots.

The Canvas logo still looks like a coronavirus to me.

23.07.2025 23:01 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Logo consisting of "Instructure" with a red period at the end

Logo consisting of "Instructure" with a red period at the end

The OpenAI integration is not the only thing Instructure announced today. Here is their new logo.

"The updated identity signals a modern, dynamic and future-ready approach designed to advance learning. Instructure is setting potential in motion."

www.instructure.com/press-releas...

23.07.2025 23:01 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 1    πŸ“Œ 1
Preview
Global hack on Microsoft product hits U.S., state agencies, researchers say Unknown attackers exploited a β€œsignificant vulnerability” in Microsoft’s SharePoint collaboration software, hitting targets around the world.

Yikes for higher education in Canada, where the past 5 years have been spent going all-in on Microsoft products. wapo.st/3TRrqcP

20.07.2025 23:13 β€” πŸ‘ 5    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
UBC discourages its privacy impact assessments’ inclusion on hub created by former staffer The Canadian Privacy Library is a new centralized website to find the privacy impact assessments (PIA) from universities across BC.

I spoke with @ubyssey.ca about the Canadian Privacy Library and UBC’s opposition to the project. UBC is following in Proctorio’s footsteps, and you’ll see how. ubyssey.ca/news/ubc-dis...

18.07.2025 18:18 β€” πŸ‘ 4    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0
AI and academic misconduct - CS072506 An MBA student complained about the feedback they received for their dissertation, saying that it was too generic to be useful. They said that an AI detection tool showed that the feedback had been pa...

Office of the Independent Adjudicator Case CS072506:

"[A student] said that an AI detection tool showed that feedback had been partly AI generated."

The student lost, because "AI detection tools can be unreliable."

www.oiahe.org.uk/resources-an...

15.07.2025 23:20 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Office of the Independent Adjudicator Case CS072505:

"A postgraduate student was invited to respond to allegations of suspected academic misconduct after Turnitin indicated that an essay contained a high proportion of AI-generated content, including hallucinated references."

The student settled.

15.07.2025 23:17 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
AI and academic misconduct - CS072504 An international student was asked to attend a viva because Turnitin had indicated that their module assignment contained a high percentage of AI-generated content. The provider was not satisfied with...

Office of the Independent Adjudicator Case CS072504:

"An international student was asked to attend a viva because Turnitin had indicated that their module assignment contained a high percentage of AI-generated content."

The student won partial justification.

www.oiahe.org.uk/resources-an...

15.07.2025 23:16 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
AI and academic misconduct - CS072502 An international student was invited to attend a viva because Turnitin had identified that their coursework contained substantial amounts of AI-generated content.

Office of the Independent Adjudicator Case CS072502:

"An international student was invited to attend a viva because Turnitin had identified that their coursework contained substantial amounts of AI-generated content."

The student won.

www.oiahe.org.uk/resources-an...

15.07.2025 23:15 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
AI and academic misconduct - CS072501 A student with autism was invited to respond to allegations that they had included large sections of AI-generated content in an essay submission.

Office of the Independent Adjudicator Case CS072501:

"A student with autism was invited to respond to allegations that they had included large sections of AI-generated content in an essay submission."

The student won.

www.oiahe.org.uk/resources-an...

15.07.2025 23:14 β€” πŸ‘ 3    πŸ” 0    πŸ’¬ 0    πŸ“Œ 1
Preview
Students win plagiarism appeals over generative AI detection tool Ombudsman tells universities to be mindful ofΒ β€˜limitations’ of detection tools and to consider if they are biased against international students

Here is additional coverage from Times Higher Education, which doesn't link to the Office of the Independent Adjudicator's case summary for some reason.

www.timeshighereducation.com/news/student...

15.07.2025 21:41 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
Casework note: Complaints relating to AI and academic misconduct This casework note discusses some of the issues and learning points we see arising from complaints relating to AI and academic misconduct.

Students won against Turnitin's faulty AI detector.

The Office of the Independent Adjudicator specifically cautioned schools from relying on AI detectors which "could be biased against a student's writing style... or if English is not the student's first language"

www.oiahe.org.uk/resources-an...

15.07.2025 21:40 β€” πŸ‘ 6    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0

Listen to students.

15.07.2025 19:18 β€” πŸ‘ 13    πŸ” 6    πŸ’¬ 2    πŸ“Œ 0
2025 Conference β€” Civics of Technology

If you thought the free, online @civicsoftech.bsky.social conference keynotes - Audrey Watters and Chris Gilliard - were πŸ”₯πŸ”₯πŸ”₯, you should look at the incredible sessions occurring on July 31 and August 1. Love to see all the ways folks embraced the theme "Communal Resistance to Artificial Systems".

06.07.2025 22:57 β€” πŸ‘ 34    πŸ” 13    πŸ’¬ 2    πŸ“Œ 2
Preview
California colleges spend millions to catch plagiarism and AI. Is the faulty tech worth it? – The Markup Colleges and universities renew Turnitin subscriptions year after year even though its flawed detectors are expensive and require students to let the company keep their papers forever.

The Markup analyzed millions of dollars worth of Turnitin purchases and asked teachers: is it worth it? themarkup.org/artificial-i...

26.06.2025 17:27 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Instructure Data Privacy Litigation - EdTech Law Center K-12 students and their parents sued Instructure, Inc., a learning management system, alleging that Instructure generates, collects, uses, and shares information about them without proper consent. Pla...

Instructure, the company that makes Canvas, is being sued for monetizing student data.

Complaint: storage.courtlistener.com/recap/gov.us...
Docket: www.courtlistener.com/docket/69805...
I wrote about this unethical practice in 2020: linkletter.org/instructure-...

25.06.2025 19:58 β€” πŸ‘ 8    πŸ” 1    πŸ’¬ 1    πŸ“Œ 1
How Instructure’s leadership monetizes student data to maximize shareholder value | Stand Against Proctorio's SLAPP!

In 2020 I wrote about Instructure's monetization of student data (linkletter.org/instructure-...). The company has been sold now and I am no longer up to speed with their practices, but I note they were sued earlier this year for doing it: www.classaction.org/news/class-a...

25.06.2025 19:42 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@linkletter.org is following 20 prominent accounts