Data Centres are Raising Electric Bills and NYC AI Chatbot Issues Illegal Advice (Issue 39, 2024)
Also, Facial Recognition Execs will Hopefully be Held Liable and You Should Be Careful of Aerial Surveillance by your Homeowner's Insurer
We’re delivering you the hottest internet news that affects all of us. Scroll down to read our full reporting below and if you love what we’re doing, please consider contributing $5 per month so that we can continue providing you with this vital, unique reporting.
Talk Liberation is committed to providing equal access for individuals with disabilities. To view an accessible version of this article, click here.
In this edition:
Facial Recognition Executives may be Personally Liable
SEO Leak Showcases the Power of Google Search
Further Experiments in Anthropomorphizing Computer Code as Persons
Big Tech Data Centres are Driving Up Electric Bills
State Senate Overrides Governor’s Veto on CBDC
OpenAI GPT-4o Model Clones User Voices without Permission
Aerial Surveillance Leads to Homeowners Losing Insurance
Meta Resumes Training AI on Your Public Images
Facial Recognition Executives may be Personally Liable
Clearview AI is a US-based facial recognition startup that previously made headlines for scraping 30 billion personal images from the web to create a searchable database. Due to the lack of consent by Europeans who were included in this database, the Netherlands’ data protection authority, Autoriteit Persoonsgegevens (AP), has penalised Clearview AI €30.5M for their violations of the GDPR.
Clearview’s response has been to claim that the EU has no jurisdiction over them, with their chief legal officer Jack Mulcaire declaring, “Clearview AI does not have a place of business in the Netherlands or the EU, it does not have any customers in the Netherlands or the EU, and does not undertake any activities that would otherwise mean it is subject to the GDPR.”
To date, Clearview has taken zero action to remedy the violations and has amassed nearly €100M in unpaid privacy fines from the European Union. Dutch DPA chairman, Aleid Wolfsen, is ready to try a different approach to hold Clearview accountable. According to Wolfsen, “We are now going to investigate if we can hold the management of the company personally liable and fine them for directing those violations.”
SEO Leak Showcases the Power of Google Search
Search Engine Optimisation (SEO) is a $74B industry that is primarily dominated by Google. A recent leak in May of 2024 revealed how Google may choose to promote or demote content and confirmed the existence of “twiddlers”, or re-ranking algorithms, and the tracking of new, previously unknown data points.
While it was widely suspected that twiddlers played a role in how Google structured and defined the internet for us, the lack of information on what determines the final search result is concerning. Additionally, the SEO leak lent evidence that smaller websites are being penalised, even when their content is of higher quality compared to content published under brand names (who are spending lots of advertising dollars).
Among the tracked data points included in the leak were entries for isCovidLocalAuthority and isElectionAuthority, appearing to indicate an internal Google process for determining which web pages appear in your searches related to the Covid pandemic and elections. With over 90% of searches in the United States conducted via Google, this outsized influence over the information we see would be less worrying if there was better transparency into how their search formula operates.
Further Experiments in Anthropomorphizing Computer Code as Persons
In October of 2023, the New York City Office of the Mayor announced that improvements would be brought to government and the community by artificial intelligence and introduced "MyCity Chatbot" which would inform on starting and operating a business in the city.
In March of 2024, The Markup found that - over the course of five months - the Microsoft Azure AI-powered chatbot was telling businesses to break the law. The Institute of Chartered Accountants in England and Wales, established by a human being via formal written orders in 1880, observed in advice to its membership of UK Chartered accountants that "any use of generative AI should be approached with caution and critical thinking. Members should engage their professional scepticism when using AIs.
Part of the issue is the confidence in which AIs can present incorrect data, which can lead people to trust what it’s giving them “like an ill-informed but confident junior”. Meanwhile, a number of corporate boards had declared computer code "entities" and have added such to their board of directors.
Big Tech Data Centres are Driving Up Electric Bills
Utilising non-disclosure agreements with local government leaders, Amazon, Meta, Microsoft and others have been quietly building massive data centres around the United States. Due to these hushed agreements, it has been very hard for local citizens to gather any information on the proposed data centre projects. This has minimised objections and left many living near noisy, 1000+ acre server farms they had no idea were going to be built.
Meanwhile, the public is shocked to learn that the U.S. government has allowed Microsoft to restart the Three Mile Island nuclear reactor to power an "AI" data centre without consultation via a $1.6 billion, 20 year contract to Constellation Energy. In an event hosted by the Bipartisan Policy Center, Jensen Huang, the founder and CEO of NVIDIA, who is also involved in the new Global AI Infrastructure Investment Partnership (GAIIP), congratulated Constellation Energy on their announcement of the recommissioning of the Three Mile Island Unit 1 reactor.
In Georgia, the demand for energy caused by more than 50 data centres raised residential utility bills by nearly $200 per year on average. This prompted state senators to recently pass a bill pausing tax incentives for developing new data centres over the next two years.
Currently there are no regulations in place to curb what some have described as the “mercenary tactics” used by the companies building the data centres. This has left residents and environmental groups with little support when trying to obtain basic information about the impacts of these massive industrial projects on energy prices, water use and noise.
State Senate Overrides Governor’s Veto on CBDC
In early September, the North Carolina State Senate overturned a veto by Governor Roy Cooper, thus passing a bill severely limiting the impact of a Central Bank Digital Currency (CBDC). House Bill 690 prohibits any state governmental agency from accepting payment using CBDC and also prohibits them from participating in the testing of CBDC.
With the passing of this bill, North Carolina has become the 6th state taking steps to block the use of CBDC, as they join Indiana, Florida, South Dakota, Tennessee, and Utah.
At the heart of this preemptive action against CBDC is the issue of federal government control. If a CBDC replaces cash, as many fear it may, the level of financial oversight afforded to the government is nearly complete. According to Bloomberg, the digital yuan pilot CBDC rolled out in China, “offers China’s authorities a degree of control never possible with physical money”.
OpenAI GPT-4o Model Clones User Voices without Permission
During August of this year OpenAI released data on the limitations and safety testing procedures for their new GPT-4o model. Included in the document was information on how the model’s “Advanced Voice Mode unintentionally imitated users' voices without permission”.
The GPT-4o model offers the Advanced Voice Mode feature so that users may utilise their voice for conversations and provide commands to the AI assistant. This is possible due to the model’s capability to imitate any voice based on a short audio clip, and apparently “synthesize almost any type of sound found in its training data, including sound effects and music”, according to OpenAI.
Downplaying the risks posed when, “a noisy input somehow prompted the model to suddenly imitate the user's voice”, OpenAI is claiming their safeguards will prevent any unauthorised voice generation. But, even if GPT-4o won’t allow the cloning of other human’s voices, according to independent AI researcher Simon Willison, “there will be models that do this that we can run on our own machines sometime within the next year or so."
Aerial Surveillance Leads to Homeowners Losing Insurance
A former Farmer’s Insurance agent has recently quit the agency in protest over its surveillance policies. The agent told the Wall Street Journal the insurer was, “dropping customers over aerial images that were 2 or 3 years old, and in one case even flagged a house for overhanging tree branches that turned out to be shadows”.
Satellite footage and drones are increasingly being used by home insurance companies to determine the state of properties they insure and, along with AI, aid in making decisions on whether to raise your premiums or refuse to renew an existing policy.
This is prompting privacy advocates like Albert Fox Cahn, founder of the Surveillance Technology Oversight Project, to call for an update to insurance regulations. According to Cahn the lack of regulations is hurting homeowners because, “State law hasn’t caught up with the technology” prompting a “need [for] updated court rules for the AI age”.
Meta Resumes Training AI on Your Public Images
Meta has announced that it will resume training its AI models on public images of adults in the UK who use Instagram and Facebook. Previously, regulatory backlash had caused the social media giant to pause the training across the EU in June of this year.
Britain's Information Commissioner's Office (ICO) had made requests for information on the processes Meta intended to use over the summer, likely following privacy watchdog NOYB’s scathing critique that the original practices lacked sufficient transparency and privacy preservation.
The primary changes implemented by Meta will now allow users to more easily object to the processing of their data along with lengthening the window of time to file an objection. Users of Instagram and Facebook in the UK began seeing the notifications to opt-out of the AI training starting on September 20th.
That concludes this edition of Your Worldwide INTERNET REPORT!
Remember to SUBSCRIBE and spread the word about this unique news service.
This issue of Your Worldwide INTERNET REPORT was produced by the Talk Liberation Digital Media Team.
Talk Liberation - Your Worldwide INTERNET REPORT was brought to you by Panquake.com. We Don’t Hope, We Build!
© Talk Liberation Limited. The original content of this article is licensed under a Creative Commons Attribution-ShareAlike 4.0 International license. Please attribute copies of this work to “Talk Liberation” or talkliberation.com. Some of the work(s) that this program incorporates may be separately licensed. For further information or additional permissions, contact licensing@talkliberation.com