by Greg Piper
Google’s artificial intelligence isn’t particularly bright when it comes to evaluating publishers’ compliance with its advertising policies, if the experience of a heterodox economics blog with outsized influence is any indication.
With a megaphone from Twitter Files journalist Matt Taibbi, both darlings of progressives in the “Occupy Wall Street” era, Naked Capitalism accused Google of making “flagrant errors” in its threats to demonetize the 18-year-old site for verboten content.
The Alphabet-owned company made censorship demands that were “not advertiser-driven,” listed “non-existent posts in the spreadsheet” shown Naked Capitalism, and issued “wildly inaccurate negative classifications” of its content, according to editor “Yves Smith,” the pen name of Aurora Advisors founder Susan Webber.
Big Tech, advertisers and activists also face growing scrutiny from Congress, state attorneys general and courts for allegedly squelching viewpoint diversity by seeking to starve contrarian publishers of ad revenue.
The House Judiciary Committee’s Weaponization of the Federal Government Subcommittee held a hearin
The committee is also saber-rattling against the World Federation of Advertisers and its Global Alliance for Responsible Media, sending warnings Wednesday to its five-member “steer team” that they may be violating “U.S. antitrust laws and congressional intent by coordinating GARM members’ efforts to demonetize and eliminate disfavored content online.”
GARM’s coordination “does not always revolve around ‘brand safety’ and ‘harmful’ content as GARM publicly claims” but also ideological motivations, committee Chairman Rep. Jim Jordan said, citing information already given the committee.
He asked the CEOs of Diageo, GroupM, Mars, Procter & Gamble and Unilever for more than five years of documents and communications “referring or relating to the categorization, demonetization, or elimination of online speech” and “conservative media outlets, including Fox News, Daily Wire and Breitbart.”
Jordan wants the same regarding “advertising, so-called misinformation, so-called harmful content, or disfavored content” on Joe Rogan’s podcast, X and changes to the social media platform after Elon Musk’s acquisition, and each company’s involvement in GARM, including “call notes of discussions or meetings.”
Neither GARM nor Google answered queries Thursday.
What we’re asking for:
cc: @FoxNews, @DailyWireNews, @BreitbartNews, @joerogan, @elonmusk. pic.twitter.com/GubxFB17ax
— House Judiciary GOP (@JudiciaryGOP) March 27, 2024
Missouri Attorney General Andrew Bailey, who with Louisiana is suing the feds for coercing social media to censor certain narratives, sued Media Matters for America for refusing to cooperate with his investigation of its allegedly fraudulent campaign to scare advertisers away from X, as it did with a similar Texas investigation.
Musk’s platform is also suing MMFA for manipulating its algorithms to pair “pro-Nazi content” with major advertisers to drive them away. The most recent activity in the docket is MMFA’s March 11 request to stay legal discovery, which could expose unflattering behavior by MMFA, pending a ruling on its motion to dismiss.
X got bad news Monday, however, when a court dismissed its lawsuit against the Center for Countering Digital Hate for “unlawfully” scraping its data to create “flawed” research in a bid to drive away advertisers. Musk’s company said it would appeal.
Biden administration counterterrorism officials sought counsel from U.K.-based CCDH, which also has a U.S.-registered nonprofit, according to a recent Freedom of Information Act production.
If there has been any attempt to defraud Missourians in order to trample on their free speech rights, I will root it out. https://t.co/QnfEMy8Mr0
— Attorney General Andrew Bailey (@AGAndrewBailey) March 27, 2024
Naked Capitalism shared the email warning and screenshots from the spreadsheet it received from its unidentified ad service, citing several Google policy “strikes,” in a March 5 post. (Smith credited the spreadsheet to Google because the ad service “regularly deals with very detailed ad placement/ revenue information in spreadsheet form.”)
“If Google identifies the flags consistently and if the content is not fixed, then the ads will be disabled completely to serve on the site,” the email reads, citing policy violations for violent extremism, hateful content, harmful health claims and “anti-vaccination.”
Those last three got paired with another strike – “demonstrably false democratic process” – for a cross-post about a “mild critic of foreign policy,” Smith said.
“Google’s demand is capricious, arbitrary, and demonstrably false” because the post “has nothing whatsoever to do with health or health care policy” and Naked Capitalism “is not vaccine hostile,” she wrote.
Indeed, a March 24 post by a different writer touts a “director’s cut” of links showing the ongoing threat from COVID-19, benefits of vaccines and former Surgeon General Jerome Adams as a “beacon of sanity” for encouraging people to wear masks as “a social cue or reassurance” for others “to do the same.”
“Google has blatantly mislabeled unrelated content to try to make that bogus charge” and Smith is consulting her lawyer, she wrote.
Fascinating how wearing masks (or not) can be contagious.
When boarding my plane, I had on my mask. Several people upon noticing me, also retrieved theirs and put them on. Seems seeing others wearing masks provides a social cue or reassurance to do the same. 😷👍🏽 pic.twitter.com/ywEGgGGfZB
— Jerome Adams (@JeromeAdamsMD) March 1, 2024
Taibbi praised Naked Capitalism as a rarity in financial journalism, not staffed by reporters “who’d jump at a job offer from the companies they cover,” and for prompting high-level Securities & Exchange Commission and California Pension Fund resignations.
“Whether it’s AI or just an algorithm … the decision to even once tell a media outlet that it must remove content to remain ad-eligible sends a message someone like Smith will think of every time she publishes going forward,” he wrote. (EWeek describes algorithms as “the essential building blocks that make up artificial intelligence.”)
Smith followed up March 21 with more details on the flagged posts and expert advice she had received that “Google relied on algorithms to single out these posts.”
It did not “accurately identify unique posts that had advertising on them” – 14, not 16 – flagged 10 posts from 2022 or earlier, only gave policy explanations for eight and singled out six that were “news summaries.” One is a topic page that cites “anti-trust actions and lawsuits” against Google, suggesting retaliatory flagging.
Google seemed to be particularly confused about the four-strike cross-post that mentioned “demonstrably false democratic process,” a review of the Chalmers Johnson book “Blowback” about the consequences of American empire, Smith said.
It has one plausible reference to health or vaccines – “When I tell my Democrat friends that Biden’s COVID policies have been arguably worse than Trump’s, they are shocked that I would say such a thing” – and refers to former President Trump in the fragment “his insistence that the 2020 election was ‘fake’ or ‘rigged,'” which Google apparently mistook for the writer’s view.
A flagged comment included a link to an article about the Biden administration’s failure to get 13 amendments to “international health regulations” through the World Health Organization.
Another, by a Barnard College economist, judged a Robert F. Kennedy Jr. claim as “false … but not outrageous” – that any reduction in death from COVID vaccines is “outweighed by an increased risk of death from other factors.”
Two news summaries were flagged, which linked to a Journal of the American Medical Association paper on “waning … vaccine-induced immunity” against COVID and criticized vaccine mandates while promoting “non-pharmaceutical” mandates.
New paper
Effectiveness of COVID vaccine primary series against symptomatic disease from omicron is nearly 0% by 9 months.
With a booster, it is about 10% effective by 9 months.https://t.co/ueMb2SHQOn pic.twitter.com/fecs94yn5Y
— Kevin Bass PhD MS (@kevinnbass) May 4, 2023
Another publisher “whose beats overlap considerably with ours” told Naked Capitalism it got a similar warning, albeit without the demonetization threat, but convinced Google to back down on claims that were “erroneous” and explain that Google’s other flags pertained to “comments, not the posts proper.”
“We have never gotten any complaints from readers or from our ad agency on behalf of advertisers” about any posts, and Naked Capitalism makes “a large budget commitment to moderation” relative to its size, such that a Columbia School of Journalism course has a session on its comments section, Smith wrote.
Yet it now faces “an ad embargo over posts that at best are less than 0.1% of our total content,” she said.
– – –
Greg Piper has covered law and policy for nearly two decades, with a focus on tech companies, civil liberties and higher education.