The Silicon Valley Bank crisis shows how panic spreads. Companies like Alethea and Graphika are using AI software to fight disinformation.
(Bloomberg Markets) — Shawn Eib was on the case as soon as customers started pulling billions of dollars from Silicon Valley Bank. At home in Virginia, he began scrolling on his laptop through the internet’s dark corners. Dubious sources, including a website the US military accuses of promoting Russian propaganda, fanned the panic.
Another posting, which prosecutors have tied to a Chinese business executive facing US fraud charges, blamed the bank’s failure on the Communist Party of China’s “unrestricted financial warfare.” Many others paired comments with come-ons for cryptocurrency get-rich-quick schemes. “I could have made a fortune like this guy!” one read.
Eib, a 38-year-old former IT administrator at Loyola University New Orleans, is an online disinformation detective for hire. He works for Alethea Group Inc., one of a handful of cybersecurity startups that wade through the online muck on behalf of companies, nonprofits and governments that want to protect themselves from internet lies and social media manipulation.
Venture capital firms and private investors committed more than $300 million to these kinds of startups from 2018 to 2022, according to data company PitchBook. Last year alone, Microsoft bought Miburo, founded by counter-extremism researcher Clint Watts, who now leads its Digital Threat Analysis Center; Spotify purchased another extremism monitor, Kinzen; and Amazon.com was among those that invested a combined $24 million in Logically, which tracks digital hate speech, foreign propaganda and crypto scams.
The SVB crisis illustrates why there’s such interest. Built over years or decades, an organization’s reputation represents intangible wealth that can make up much of its value. In the internet era, it can be destroyed overnight. The broader issue would be outright lies, not just posts that fan panic, that undermine the world’s financial system. The European Banking Authority has warned that banks are “vulnerable” to runs because of “campaigns that spread inaccurate information.”
To address this risk, the anti-disinformation startups typically have artificial intelligence software that scours the web for suspicious content. They then issue reports or alert corporate victims about online lies. The companies make money in a variety of ways. Some charge consulting fees or work on retainer. Alethea declined to discuss pricing. One of its main competitors, Graphika Inc., says typical arrangements are subscriptions starting at roughly $100,000 a year.
The bank run also shows the challenges and limitations of what’s come to be known as the disinformation mitigation industry. After all, SVB couldn’t be protected, and it was difficult to separate malefactors from those legitimately pointing out the true risk of the bank’s poor management decisions.
It’s also hard for outsiders to verify claims about the industry’s effectiveness. Samuel Woolley, who researches propaganda at the University of Texas at Austin, says these companies are “thin on details” when discussing their methods. “It’s easy to be technologically deterministic,” says Woolley, an assistant professor who says he’s declined offers to create his own disinformation detection company. “But you have to take into account social and cultural context—like the management of a bank.”
Washington-based Alethea and others acknowledge their shortcomings and say they must shield their methods from wrongdoers. They say their services can be a hard sell. Companies don’t tend to have “disinformation budgets,” says Ted Schlein, general partner and chairman of Ballistic Ventures, which backs Alethea. Lisa Kaplan, Alethea’s founder and chief executive officer, recalls how pitching her services has felt more like “selling health insurance to smokers in the ’70s.”
In 2022, four-year-old Alethea raised $10 million from Ballistic, whose partners have worked in cybersecurity for the military, US intelligence agencies and AT&T Inc., as well as at VC giant Kleiner Perkins. Alethea is a woman’s name derived from the Greek word for truth. Along with Kaplan, a former digital director for Maine Senator Angus King’s 2018 reelection campaign, the rest of its top leadership comprises women.
Alethea issues free reports about threats to journalists and other targets; that helps it win over paying customers. Clues that posts could be fakes or controlled by someone else might be a Twitter account with 50,000 followers consistently getting 15,000 likes, a tweet retweeted more than it’s been liked, or websites sharing the same domain address. Alethea recently warned an American telecom company, which it declined to name, about “Spamouflage Dragon.” Posing as customers, the network of China-linked accounts was spreading disinformation.
Graphika in New York has a longer track record. John Kelly, a computational social scientist with a doctorate from Columbia University, founded the company in 2013. Graphika says its clients include Facebook parent Meta Platforms Inc. and Alphabet Inc.’s Google, and it’s also worked with the US Senate Select Committee on Intelligence.
In the 2022 US midterm elections, Graphika linked the Russian Internet Research Agency—which the US has charged with meddling in the 2016 elections—to accounts impersonating the singer Kid Rock to promote conspiracy theories. It’s also uncovered an Egyptian company running fake news pages aimed at Ethiopia, Sudan and Turkey and ersatz clusters of accounts promoting Chinese telecommunications giant Huawei Technologies Co., which faces US sanctions.
In the case of the recent banking chaos, Graphika’s Santiago Lakatos, who tracks finance and markets, used the company’s software to examine what he calls the “extensive map of conspiratorial finance communities.” Like Alethea, he suspected adversaries would use the crisis to destabilize the US. He also found cryptocurrency advocates fueling bank runs not only at SVB but also at New York-based Signature Bank and San Francisco-based First Republic. Along with “piling in on SVB,” Russian state actors over the past 24 months have been sparking discontent over inflation and the cost of living in the US and Europe, says Jack Stubbs, Graphika’s lead analyst.
Of course, one person’s disinformation can be another’s statement of political views, so the companies have to walk a fine line. Graphika provides intelligence to organizations that, the company says, can make their own strategic decisions. “We aren’t trying to be the arbiters of what is true or false,” says Guyte McCord, Graphika’s chief operating officer.
Red flags can be misleading, too. Michael Robinson, Alethea’s lead analyst, investigated some Japanese users when he worked at Twitter Inc. They had a suspiciously large number of accounts, suggesting a fake network. The users, it turned out, were using them as different personas in online role-playing games.
There’s also the question of what can be done. David Simons, a partner at law firm Boies Schiller Flexner, says it’s often hard to take legal action because internet platforms aren’t liable for their content and the malefactors are difficult to pinpoint. Mitigating lies can be too costly for some organizations. “It’s much harder for smaller companies to fight back,” he says.
McCord compares the SVB bank run to the meme stock craze of 2021, when retail investors bid up the stocks of companies such as GameStop Corp. and AMC Entertainment Holdings Inc. to absurd levels, in part to squeeze Wall Street investors who’d bet against the shares. The latter led to devastating investment losses, ultimately for those on both sides; the former destabilized the banking system. The two examples sum up the thorny problem. On the web, posts can start out true (or close to it), morph into lies and panic, and have consequences few could anticipate.
Murphy covers cybersecurity for Bloomberg News in San Francisco.
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.