Sunday, March 05, 2023

The "most fabulous conference @BruxConf2023", unrolled

Historic times that we are living right now @vestager 
A *very* long chapter on subsidies, understandably - entrepreneurs, not taxpayers, creating jobs...#BruxConf2023 
Green claims - calling the green bluff if we see it @vestager #BruxConf2023 
Digital markets have not fulfilled their promise(s)..."we cannot be accused of having been too fast" @vestager #BruxConf2023 
"Reformulating concerns" - direct focus on consumers themselves #Apple 
"Full toolbox" 🗜️🔧🔨✂️🔭🪛🪚🔮 
International cooperation in merger control important indeed - the @WSJ had an interesting article on that, recently... 
"We have war in Europe"@vestager top of our mind 
Political economy, the idea of cheap is going away @RanaForoohar #BruxConf2023 
One of my personal heroes 👉@chopracfpb 
Promoting entry, promoting resilience, a way of *American/Germanic* thinking? "It's a new book that we are writing" #BruxConf2023 
[we had the "new book" discussion only yesterday evening with a dear friend economist] 
Antitrust is the bare minimum - open banking's plug, sector regulation @chopracfpb 
"We all know why" US Big Tech legislation wasn't passed @chopracfpb - the EU #DMAMiracle instead - not perfect but workable 
Yep, new economic paradigm not very clear yet - exciting times! @RanaForoohar 
"It's settled a bit that it didn't work" @chopracfpb 
"This isn't the coffee break" @Caffar3Cristina makes clear 
Let's discuss the Biden Executive Order - Really @superwuster #BruxConf2023 
[Pity I wrote this concurrences.com/IMG/pdf/06.con… too early]
"It's an expert thing" - well, I think that effects of the shift should also be palpable (kitchen table topic, e.g. "look, we have interoperability we grandma's favourite service!"] 
Exculpatory narratives (@superwuster), demand for them and agencies believing them - what could possibly go wrong? 
In full praise of the New Brandeisians, apparently @AMundt_BKartA #BruxConf2023 
But what bout the Courts? @AMundt_BKartA #posture v. #law 
"Not close to the more economic approach" but revising the economic toolbox @AMundt_BKartA #BruxConf2023 
...with a little help from the German legislator - and more coming 😉#BruxConf2023 
Thrilled that many are mentioning UK open banking - it's much about interoperability (open APIs, etc.), guys #BruxConf2023 
More economics, less enforcement, suboptimal economic effects @TomValletti - one merger per year prohibited by @EU_Competition 
The "Orthodox but progressive" economist says, please, not more "allegedly sophisticated economics" - incentives, resources all wrong @TomValletti - simple and coherent economics needed 
...economics helpful in understanding business strategy, incentives, etc. Agree, but well beyond IO, if I may 
The amazing @johnmarknewman up now: "the law has some catching up to do" 
Rusty legal tools - but not dead law #Meta_Within - they are on the table, and sharp enough 
Political economy we don't want: companies as central planners ("buying instead of building") @johnmarknewman 
"Antitrust is failing" and this hurts @rodgsims #BruxConf2023 
More structural provisions in merger control and more per se rules @rodgsims 
"Empire strikes back" moment might be coming @superwuster 
Common law nature of some of the changes
@superwuster
- [guidelines and other means can help institutionalize them, of course] 
"Data inflation" - legislators didn't see it @AMundt_BKartA 
Discussions, postures ok, but achieving real change is plan A - Andrea Coscelli 🙌 #BruxConf2023 
Not a 'Chicago style' simple theory is what we need, but a set of strong, flexible principles @johnmarknewman #BruxConf2023 
Ristretto and some fresh air urgently needed - but no Belgian patisserie hereImage
"Non-compete" should be a huge red flag @RKSlaughterFTC #BruxConf2023 
A very healthy industry focusing on suing the FTC @RKSlaughterFTC - not unexpected 
A history teacher before going to law school, Doha Mekki #BruxConf2023 - very useful in the current debate 
AI/ML - how is the role of data changing? Doha Mekki 
Decarbonization changing affected industries, but how? @BCoeure #BruxConf2023 
Plugging the legal gaps: more international cooperation in antitrust needed - Margarida Matos Rita #BruxConf2023 
Now the orthodox and progressive Martijin Snoep - also national *regulator* under the DMA (nice he mentioned it) #BruxConf2023 
Nice case before EU courts in merger control just mentioned "how significant" should be each individual transaction be if there are many? 
[New Competition Tool off the table - still mourning myself] 
Fairness and exploitative abuses - current Apple cases moving in that direction, we are hearing #BruxConf2023 
More sustainable supply chains in order to avoid (?) the climate disaster - Article 101.3 comes to rescue, with a couple of nudges (4.000.000 🐄in the NL) 
"Whole-of-government approach" beyond the FTC/DOJ [some agencies dragging their feet, however - my impression] #BruxConf2023 
A Cross-Atlantic chicken problem discussed like now, great! - #kitchentabletopic #BruxConf2023 
"Don't move, this isn't lunch time yet" @Caffar3Cristina 
"Courts have considerable leeway" AG Kokott #BruxConf2023 
1) Containing market power, corporate liability extended;
2) Strengthening defence rights of companies of market power - more economic approach still blossoming in the EU - perhaps revising EU Art. 102 guidelines might have helped; 
[Worrying, from a Judge: more economic approach here to stay, underenforcement a real risk] 
3) Killer acquisitions: plugging a gap possible 🙌 
Supplementary application of Art.102 (2), Art. 22 MR...Illumina good news so far, of course. 
DMA: gaps plugging in the area of merger control? 🤔#BruxConf2023 
*(2)* was actually (?) - very sceptical myself 😊 
Jonathan Kanter #BruxConf2023, all in praise of @Caffar3Cristina 👏👏👏 
"We are just getting started" @JusticeATR 
Digital:
1) How does competition in this industry present itself? A tough but essential Q. "We need models that fit facts":
-intermediaries, often endless possibilities of anticompetitive behaviour
-- concentration, tipping markets (interrelated actions)
-- setting the rules 
Self-preferencing label, not a safe-harbour Jonathan Kanter - #BruxConf2023 
Structural reliefs always on the table (thus, e.g., "incentives to interoperate" may also be triggered 👍), as already shown -BTW 
"An all-economist panel" - haven't we just heard that the more economic approach has caused underenforcement 🤔? Reflection on what went wrong but able to come up with something better, now? #BruxConf2023 
Models getting it wrong - because, e.g., free entry assumptions are, well, fugazi - nobody to blame? 
Innovation and theories of harm @johnvanreenen (me: not waiting for IO economists to develop them); praising ex-ante regulation (also in order to incrementally build expertise, an important point made by Chiara @Arcep on Monday) 
Yes, gatekeepers are going to drag their feet...So what 😊? The DMA is powerful (it helps if economists might read all those complex provisions) 
Overinvesting ("in buying") in order to keep out the competition @jan_eeckhout 
Network theory of market power - nodes and links @BPellegrino_UMD - no need to define markets (well, this is becoming less important in any case 😉) #BruxConf2023 
Working in the @FTC trenches, Aviv Nevo - we need "more economics in terms of being broader" #BruxConf2023 
Transaction values matter, says Chiara Fumagalli. Well, the DMA at least is shining some light on all the transactions by gatekeepers - focusing on the high-value ones as a second step, perhaps -#BruxConf2023 
Upcoming Merger Guidelines @Susan_Athey
What's competition, in this setting? Increasing barriers to entry, blocking a nascent competitor, future competition, and so on... 
Thinking about market definition "more realistically", to avoid "confusion", and deriving the relevant market directly from the facts @Susan_Athey #BruxConf2023 [actually, at least in Meta/Within, it was actually not much of a burden] 
Pierre Régibeau - killer acquisitions below the threshold are a thing #BruxConf2023 
PR still wishing we could devote some (academic, possibly) thinking to things we could be better - don't getting "crazy about" new theories of harm because, of course, we have so much time as we like because in the long term we'll be all dead in any case #BruxConf2023 OK 
Turning point: Furman Review - Sarah Cardell @CMAgovUK 
Uncertainty as dealt in the UK merger guidelines - one of my favourite reads - also the UK Court "liked" them BTW #BruxConf2023 
Economic framework we need: no labels, not much overlaid complexity, but they should be helpful in explaining how competition in reality works (commercial realities) - it's not a precise science, we might occasionally get it wrong - Brava 
[Personal note: my deep frustration as an antitrust legal scholar having to cope with totally unrealistic Chicago theories developed into a PhD on vertical restraints embracing broader economic theories - 20 y. ago] 
No worries, @RepKenBuck

Still alive and tweeting - thanks to plenty of water and 4 ristretti! 
Do we need an antitrust or already a much broader policy/regulatory discussion on generative AI? #BruxConf2023 
Trying to understand the tech issues - and then the adtech case came together @KenPaxtonTX #BruxConf2023 
Stéphanie Yon-Courtin blocked me on Twitter: I respect her choice 😶 
"Democratic convergence on what we wanted to achieve" @FiloBXL #BruxConf2023 
DMA: a new form of regulatory supervision - as written in the law. 
Platforms constantly changing the rules of the road @doctorow 
Are we celebrating the wrong type of innovation? - @ArielEzrachi 
[BTW - Art. 7 DMA is not *at all* backward looking 😎] 
Before the 🍾starts to flow: Olivier Guersent on ducks and chickens #BruxConf2023 
A new generation of Article 102 cases against the background of the DMA - not the end of antitrust - (agreed) OG 
[Wrong remedies in the past, shouldn't we start taking some risks as done in the US and the UK, asks Cristina?] 
"We will do it again" 

Thursday, February 09, 2023

MISE EN ŒUVRE DU RÈGLEMENT SUR LA GOUVERNANCE DES DONNÉES (DATA GOVERNANCE ACT): LES ENJEUX ÉCONOMIQUES

 CNIL, ici

The Financial Services Sector’s Adoption of Cloud Services

 U.S. Department of the Treasury, here

Exclusive Interview: OpenAI’s Sam Altman Talks ChatGPT And How Artificial General Intelligence Can ‘Break Capitalism’

 Forbes, here

de Streel on the DMA

Chez Oles, Video here.  

Microsoft/Activision Blizzard

 CMA, Provisional findings report, here

Italy bans U.S.-based AI chatbot Replika from using personal data

 Reuters, here

“Where are we heading with digital identities?”

 EDPS, here

[Asked ChatGPT to summarize the speech: 

"The article is about the European Digital Identity Wallet framework (EDIF) proposed by the revision of the eIDAS, which aims to provide citizens with a secure and privacy-friendly digital identity. The writer, who represents data protection, believes that the EDIF has the potential to be a big success and that the relationship between EDIF and the GDPR can be examined under two different aspects. On one hand, if implemented correctly, it can effectively protect individuals and their data and comply with essential provisions of the GDPR, such as data accuracy and security of processing. On the other hand, the data processing activities related to digital identity must comply with the GDPR, particularly with data minimization and protection by design and by default. The writer expresses concern about the 28 Implementing Acts in the proposal, which will decide on the actual implementation aspects, and questions the basis for these acts and whether there are mandatory certification mechanisms in place to ensure compliance with the GDPR"]

Can you imagine your Students using ChatGPT to summarize the reading assignments instead of actually studying them? I can.

Psychologists to help the FTC evaluate alleged mental health harms

M. Bedoya, here.  

Remarks and Regrets

 FTC, here

"I don’t mean to criticize the FTC in pointing this out—I think former Commissioner Jones Harbour was incredibly visionary and ahead of her time. The FTC was just being faithful to our existing practices in merger review. But clearly there was something going on that we did not catch, given what we know about the ad tech market today. So I think it’s a real lesson learned that shows that we antitrust enforcers need to continually re-examine and fine tune our approach. The goal of our merger guideline revision project is to do just that and enable us to more effectively use our tools so we can make better predictions about the future effects of proposed mergers."

ChatGPT is a data privacy nightmare, and we ought to be concerned

 ArsTechnica, here

ChatGPT itself on the same topic:

As we live in a world where technology is rapidly advancing, artificial intelligence and machine learning models have become an integral part of our lives. One of the most popular and powerful language models developed by OpenAI is ChatGPT. It is a state-of-the-art model that has been trained on a vast amount of data and can generate human-like responses to various questions. However, while ChatGPT has many benefits, it also raises serious concerns about privacy.

First and foremost, ChatGPT is trained on a massive amount of data, including text from various sources such as books, articles, and even conversations. This data is not necessarily public or anonymous, which means that it can contain sensitive information about individuals and organizations. This sensitive information can be used for malicious purposes, such as identity theft, financial fraud, or even cyberbullying.

Furthermore, the nature of language models like ChatGPT means that it can store vast amounts of data in its memory. This data can be accessed and used by anyone who has access to the model, and there is no guarantee that it will not be misused. In fact, there have been instances where language models have been used for unethical purposes, such as generating fake news or creating biased content.

Moreover, the use of ChatGPT raises serious ethical concerns about the right to privacy. As individuals and organizations continue to share more and more personal information online, they are putting their privacy at risk. If this data falls into the hands of malicious actors, it can be used to target individuals and organizations with malicious campaigns. Furthermore, there is no way for individuals and organizations to know how their data is being used by these models, and there is no way for them to opt out of this process.

In conclusion, while ChatGPT is a powerful tool that has many benefits, it also raises serious concerns about privacy. As the use of AI and machine learning models continues to grow, it is imperative that we take the necessary steps to protect individuals' and organizations' sensitive information. This includes ensuring that these models are trained on anonymous and public data, as well as implementing strict privacy policies to prevent the misuse of sensitive information. We must also consider the ethical implications of using these models, and work towards finding a balance between their benefits and the risks they pose to privacy.

When asked about ChatGPT's compliance with the GDPR:

The General Data Protection Regulation (GDPR) is

a regulation in EU law on data protection and privacy for all individuals within the European Union (EU) and the European Economic Area (EEA). The regulation sets out the rules for how companies and organizations must handle personal data and ensure that the privacy of individuals is protected.

It is difficult to say definitively whether ChatGPT is compliant with the GDPR without a thorough review of how the model is used and how personal data is collected, processed, and stored. However, here are some considerations:

  1. Purpose Limitation: The GDPR requires that personal data be collected for specified, explicit, and legitimate purposes. If ChatGPT is used for purposes that are not compliant with the GDPR, it may not be in compliance with the regulation.
  2. Data Minimization: The GDPR requires that personal data be limited to what is necessary for the intended purpose. If ChatGPT collects more data than is necessary, it may not be in compliance with the regulation.
  3. Transparency: The GDPR requires that individuals be informed about how their personal data is being used. If ChatGPT is not transparent about how personal data is collected, processed, and stored, it may not be in compliance with the regulation.
  4. Data Security: The GDPR requires that personal data be securely processed and stored. If ChatGPT does not have appropriate security measures in place, it may not be in compliance with the regulation.

Overall, it is the responsibility of the entity using ChatGPT to ensure that the model is used in a way that is compliant with the GDPR. This includes conducting a data protection impact assessment, implementing appropriate technical and organizational measures to secure personal data, and providing individuals with clear information about how their personal data is being used.

There was a sudden system error, and then I asked again (exactly the same prompt) - getting a different answer:

As an AI model developed by OpenAI, ChatGPT is not subject to the General Data Protection Regulation (GDPR), which is a regulation in EU law on data protection and privacy for all individuals within the European Union (EU) and the European Economic Area (EEA).

However, OpenAI takes data privacy and security seriously and has implemented measures to ensure that personal data is processed in accordance with applicable privacy laws, including the GDPR. If you are using ChatGPT in the EU or the EEA, it is the responsibility of the entity that controls the use of ChatGPT to comply with the GDPR and other applicable privacy laws.

If you have concerns about the use of ChatGPT and data privacy, I would recommend contacting OpenAI directly for more information.

What happened?


Study on the provision of information to consumers about the processing of vehicle-generated data

 EC, here

Internet Centralization: What Can Standards Do?

 IETF, here

Banning Noncompetes Is Good for Innovation

 M. Lemley, O. Lobel, here

Market Study Report on Mobile OS and Mobile App Distribution

 Japan FTC, here