Kris Shrishak
krisshrishak.bsky.social
Kris Shrishak
@krisshrishak.bsky.social
Enforce Senior Fellow at @iccl.bsky.social | Tech-policy, algorithmic decision making, privacy, cryptography, Internet security, human rights | KrisShrishak@eupolicy.social
My piece on @euobserver.com following up on the letter from scientists to the President of the European Commission

euobserver.com/digital/ar68...
November 10, 2025 at 11:56 AM
You can read the letter in an image

h/t @lorak.bsky.social for the idea
November 10, 2025 at 11:10 AM
"After burning through almost a trillion dollars, we finally invented a computer that can't do math."

"For the torment nexus, that is the AI industry, to get built, it needs to convince you that your only option for survival is to help build it."
November 5, 2025 at 6:38 PM
"We are tearing people down by convincing them that they can't function without these tools. And we are doing this purely because some very rich unhappy oligarchs would like to become even richer if no more happier. There is nothing benevolent about this."
November 5, 2025 at 6:38 PM
This prohibition has applied in the EU since 2 Feb 2025. But Ireland, like other EU member states, is yet to empower the relevant regulators. This means that it is not clear which AI regulator in Ireland would be monitoring this prohibition and who could take action.
October 22, 2025 at 8:52 AM
This prohibition has applied in the EU since 2 Feb 2025. But Ireland, like other EU member states, is yet to empower the relevant regulators. This means that it is not clear which AI regulator in Ireland would be monitoring this prohibition and who could take action.
October 22, 2025 at 8:46 AM
The relevant part of the law Article 53 (4)
July 18, 2025 at 3:58 PM
Quoted in @grahamjfraser.bsky.social's article in BBC News yesterday on WhatsApp forcing people to use GenAI.
April 24, 2025 at 10:21 AM
To understand how bad the @ec.europa.eu delays are, here is a comparison from my slides comparing Commission's delay to respond to FOI requests with @deutschebahn.com .
April 15, 2025 at 8:05 AM
2. A remote attacker (and anyone who accesses the device when logged-in) can access historical information that could be used to harm people, e.g., blackmail.
April 11, 2025 at 12:32 PM
Delighted with the #privacywashing through #PETs seminar last week at Dagstuhl org w/ @carmelatroncoso.bsky.social @emilianodc.com @strufe.bsky.social

We had excellent senior researchers. We were also fortunate that young researchers who will shape the field joined us.

1/

www.dagstuhl.de/25112
March 18, 2025 at 10:27 AM
Article 26(8) does not mention data protection impact assessment. Article 26 (9) does.
March 5, 2025 at 4:20 PM
I wonder if this is a typo in the #EU #AI Act in Annex VIII Section C paragraph 5.

eur-lex.europa.eu/eli/reg/2024...
March 5, 2025 at 4:20 PM
And what happens if the "shares" are held in AWS?
February 26, 2025 at 3:17 PM
What about claims that "shares" cannot be combined?
February 26, 2025 at 3:17 PM
WorldCoin Foundation claimed that contractual provisions prevent them from joining the shares. But this argument does not hold if the two entities holding the "shares" are both data processors for the WorldCoin Foundation.
February 26, 2025 at 3:17 PM
Not always. If the "shares" are deemed as personal data (incl. pseudonymous), then GDPR applies.
February 26, 2025 at 3:17 PM
The Attorney's statement, however, seems to indicate that the training was not good sufficient.
/6
February 14, 2025 at 7:29 PM
They claim to have provided adequate training on the use of an internally developed "proprietary AI-powered legal assistance system".

storage.courtlistener.com/recap/gov.us...

4/
February 14, 2025 at 7:29 PM
In the response, the Trial Counsel and the Chief Transformation Officer place all the blame on the attorney.

storage.courtlistener.com/recap/gov.us...

3/
February 14, 2025 at 7:29 PM
The judge issued an order to show cause why the plaintiffs should not be sanctioned for citing nine cases of which only one was exists. The defendants were able to generate some of the mis-cited cases on #ChatGPT.

storage.courtlistener.com/recap/gov.us...
2/
February 14, 2025 at 7:29 PM
But they have only good intentions

"to assess the predictive power of the Photo Big 5 in labor markets, [but] not intended, and should not viewed, as advocacy for the usage of Photo Big 5 or similar technologies in labor market screening"

But what are their ethical concerns?
January 24, 2025 at 4:33 PM
"We apply this method to a large sample of LinkedIn users, focusing on MBA graduates—a high-skill and relatively homogeneous worker group"

And they claim "our methodology can be applied to any individual with a publicly available facial image."
January 24, 2025 at 4:33 PM
This is brilliant. Lets hope the European Commission will follow through on this decision from the ECJ and that harmonised standards from CEN/CENELEC (including those for #AIAct) will be free for the public.

curia.europa.eu/juris/docume...
March 5, 2024 at 11:43 AM