Miss a day, miss a lot. Subscribe to The Defender's Top News of the Day. It's free.
OpenAI’s GPT-4 Is Closed Source and Shrouded in Secrecy
OpenAI released a 98-page technical report on Tuesday to accompany its unveiling of its latest large language model, GPT-4. Among the hype surrounding the model’s new capabilities, such as its ability to pass the bar exam, is growing criticism from AI researchers who point out that the paper is not transparent or “open” in any meaningful way.
The report, whose sole author is listed as the company rather than specific researchers, explicitly says, “Given both the competitive landscape and the safety implications of large-scale models like GPT-4, this report contains no further details about the architecture (including model size), hardware, training compute, dataset construction, training method.”
This means that OpenAI did not disclose what it used to train the model or how it trained the model, including the energy costs and hardware used for it, making GPT-4 the company’s most secretive release thus far. As Motherboard has noted before, this is a complete 180 from OpenAI’s founding principles as a nonprofit, open-source entity.
AI researchers are warning about the potential consequences of withholding this information. Emily M. Bender, a Professor of Linguistics at the University of Washington, tweeted that this secrecy did not come as a surprise to her. “They are willfully ignoring the most basic risk mitigation strategies, all while proclaiming themselves to be working towards the benefit of humanity,” she tweeted.
California School Board Sues TikTok, Snap, Google Over Youth Mental Health Crisis
A California school board is suing giant social media companies for allegedly creating a “destructive environment for children” and leaving parents and educators to deal with what they deemed is a growing mental health crisis among the youth.
The 107-page lawsuit, filed Monday in Northern California on behalf of the San Mateo County Board of Education and the San Mateo County Superintendent of Schools Nancy Magee, names some of the biggest names in social media, including YouTube, Google, Snap Inc. and TikTok.
The suit alleges the companies use artificial intelligence and machine learning to deliver harmful content to children.
The lawsuit said the tech companies have knowingly created an “unprecedented mental health crisis” in pursuit of profit by purposefully designing their platforms to be addictive and to deliver harmful content to young users.
Silicon Valley Bank Collapse Produces an Early Winner: Digital Banks
The deluge of new account requests began hitting San Francisco-based Mercury last Thursday morning, the day after Silicon Valley Bank announced that it had sold $21 billion in securities at a $1.8 billion loss and needed to raise more capital.
Over the weekend, as federal bank regulators scrambled to make sure SVB’s failure didn’t set off a broader bank run, Mercury’s workers scrambled too; its normal account-opening staff was doubled to 60, as risk and compliance pros, plus volunteer software engineers and salespeople (who got a crash course in how to verify and approve new customers), pitched in.
In both cases, the extraordinary measures seem to have paid off–for now, at least. Regional banks have stabilized. And according to venture capitalists, Mercury has likely been the biggest winner so far among the fintech digital banks.
Immad Akhund, Mercury’s 38-year-old CEO and cofounder reports that in just six days his 470-person company has added more than $2 billion in deposits and thousands of customers to the 100,000 accounts it had before. He started Mercury six years ago, he says, precisely because he believed a technology-based banking platform could provide better service to startups than SVB. “I had a lot of respect for [SVB]. I used them at my previous company,’’ he says. “I’m sad about it personally. It’s been a real mix of emotions.”
Biden Admin Demands TikTok’s Chinese Owner Sell Stakes or Risk Being Banned: Official
The Biden administration is demanding TikTok’s Chinese owner sell its stake in the app or risk getting banned, the company and a U.S. official told ABC News.
TikTok confirmed to ABC News on Wednesday that it was recently contacted by the Committee on Foreign Investment in the U.S. (CFIUS). The company said CFIUS prefers for ByteDance to sell its stake in TikTok, rather than reach an agreement with the U.S. government over national security concerns.
This move is an escalation and comes as the administration has been publicly hardening its stance against TikTok. Last week, the White House came out in support of bipartisan legislation that could be used to ban TikTok.
In a statement to ABC News, TikTok spokesperson Brooke Oberwetter said: “If protecting national security is the objective, divestment doesn’t solve the problem: a change in ownership would not impose any new restrictions on data flows or access. The best way to address concerns about national security is with the transparent, U.S.-based protection of U.S. user data and systems, with robust third-party monitoring, vetting and verification, which we are already implementing.”
W.H.O. Do You Trust?
Trust in public health has been shattered — half of America has lost faith in the scientific community altogether. Apparently blind to that loss of trust and the urgent need for transparency, our current government is doubling down on World Health Organization (WHO) decrees that directly circumvent public oversight.
The WHO has drafted a new global Pandemic Accord. Before seeing anything final in what will be legally binding, the U.S. ambassador to the WHO, Pamela Hamamoto, on February 27 imprudently promised “The United States is committed to the Pandemic Accord.”
Americans should be concerned that the latest WHO agreement in progress is not classified as a treaty. America’s treaties require congressional approval. That’s key to a representative government — the public gets input into how they are governed. But executive-signed “accords” circumvent the authority of the people. After all the public has endured in this pandemic at the hands of the government, is there any doubt that health emergency powers must have full, public vetting?
Despite protestations to the contrary, all signers of the Pandemic Accord clearly relinquish critical autonomy to the WHO. Most ominous is that WHO defines “public health emergency” on its own terms — giving it full leeway to determine the fundamental justification for public restrictions. We already know there is ambiguity within the U.S. on that definition. When you’re going to invoke emergency measures, there must be clarity on the terms and time limits of the emergency. Why should any sovereign nation allow a third party to legally define and impose such a critical state?
FCC Officials Owned Stock in Comcast, Charter, AT&T and Verizon, Watchdog Says
The Federal Communications Commission should be investigated for letting employees own stock in Comcast, Charter, AT&T, and Verizon, nonprofit watchdog group Campaign Legal Center told government officials.
“Federal law specifically bans FCC employees from owning ‘any stocks, bonds, or other securities of [any company] significantly regulated by the Commission,'” the nonprofit group said last week in a letter and detailed report sent to FCC Acting Inspector General Sharon Diskin. “Despite this ban, the most recent financial disclosures publicly available show that ethics officials allowed multiple FCC employees to own stock in telecommunications and other companies that appear to fall under the prohibition.”
Citing the most recent financial disclosure reports, which cover the Chairman Ajit Pai-era years of 2018 and 2019, the Campaign Legal Center report said FCC official Rosemary Harold owned Comcast stock with a value between $3,003 and $45,000. Harold was the FCC Enforcement Bureau chief during that time and is now a deputy chief with the FCC Media Bureau. The report also said former FCC official Lisa Hone, then a deputy bureau chief, owned Charter Communications stock worth between $4,004 and $60,000.
Parents Fed up as Hundreds of Colleges Still Mandate COVID Vaccines: ‘Laboratory of Guinea Pigs’
Parents and students alike are still battling vaccine mandates at colleges nationwide more than three years after the pandemic first began.
Kristina Kristen, whose son attends the University of California-Irvine, is one of those parents. Even though her son was granted a religious exemption from the mandate, she said she is still “concerned” about the unknowns surrounding the shot.
She joined “Fox & Friends First” on Wednesday to discuss her angst surrounding the vaccine’s potential long-term side effects, and her broader scrutiny of the mandate. “They are creating a laboratory of guinea pigs by making our students take these shots without knowing the long-term effects,” she continued.
According to No College Mandates, there are at least 800 colleges that require the primary COVID vaccine, at least 250 that require a booster shot, and at least 25 that mandate a bivalent shot.
“We have to emphasize the fact that the COVID shots do not stop transmission or infection,” Kristen said. “Therefore, they serve no public health purpose, period, and this fact was known by vaccine makers very early on through their trials.”
CDC Bought Phone Data to Monitor Americans’ Compliance With Lockdowns, Contracts Show
The U.S. Centers for Disease Control and Prevention (CDC) purchased data from tracking companies to monitor compliance with lockdowns, according to contracts with the firms. The CDC paid one firm $420,000 and another $208,000. That bought access to location data from at least 55 million cellphone users.
The contracts, approved under emergency review due to the COVID-19 pandemic, were aimed at providing the CDC “with the necessary data to continue critical emergency response functions related to evaluating the impact of visits to key points of interest, stay-at-home orders, closures, re-openings and other public health communications related to mask mandate, and other merging research areas on community transmission of SARS-CoV-2,” the contracts, obtained by The Epoch Times, state.
While the data is deanonymized, it can be used to identify people, researchers have shown.
“It remains unclear why the CDC tracked millions of Americans during the pandemic and whether it continues to do so. In response to COVID-19, the CDC should have been prioritizing the development of treatments, effective testing and vaccine safety rather than tracking Americans’ daily lives,” Sen. Ron Johnson (R-Wis.) wrote to CDC Director Dr. Rochelle Walensky.
Meta’s ‘Single Largest Investment’ Is Now AI, Not the Metaverse
In stark contrast to its promise to deliver on the metaverse that led it to change its name at the end of 2021, Facebook and Instagram’s parent company Meta has announced an unsurprising shift in direction.
In a letter to employees, Meta CEO Mark Zuckerberg announced that the company’s “single largest investment” is now in advancing its AI strategy.
While artificial intelligence will undoubtedly find itself heavily integrated into the metaverse, the company’s sudden change of direction and heavy investment in AI begs the question of whether it’s seeking a short-term fix to the economic challenges that it’s very clearly facing.
OpenAI Checked to See Whether GPT-4 Could Take Over the World
As part of pre-release safety testing for its new GPT-4 AI model, launched Tuesday, OpenAI allowed an AI testing group to assess the potential risks of the model’s emergent capabilities — including “power-seeking behavior,” self-replication and self-improvement.
While the testing group found that GPT-4 was “ineffective at the autonomous replication task,” the nature of the experiments raises eye-opening questions about the safety of future AI systems.
“Novel capabilities often emerge in more powerful models,” writes OpenAI in a GPT-4 safety document published yesterday. “Some that are particularly concerning are the ability to create and act on long-term plans, to accrue power and resources (“power-seeking”), and to exhibit behavior that is increasingly ‘agentic.'” In this case, OpenAI clarifies that “agentic” isn’t necessarily meant to humanize the models or declare sentience but simply to denote the ability to accomplish independent goals.
Over the past decade, some AI researchers have raised alarms that sufficiently powerful AI models, if not properly controlled, could pose an existential threat to humanity (often called “x-risk,” for existential risk). In particular, “AI takeover” is a hypothetical future in which artificial intelligence surpasses human intelligence and becomes the dominant force on the planet. In this scenario, AI systems gain the ability to control or manipulate human behavior, resources, and institutions, usually leading to catastrophic consequences.
The U.K. Joins Other Countries in Banning TikTok From Government Devices
The U.K. has become the latest jurisdiction to ban the social video app TikTok from government devices, confirming reports that surfaced earlier this week.
The ban follows a security review ordered by U.K. ministers looking to establish whether government data could be compromised from social networks installed on work devices. Indeed, the review did include other social media apps, but the government has decided on a “precautionary ban” for TikTok only — likely because its use is limited in governmental roles, compared to something like Twitter. There is nothing to indicate that the Government has found any specific security concerns unique to TikTok.
However, the government has said that employees can still use TikTok on personal devices, while it will consider exemptions for work devices on a case-by-case basis.
