An interesting take: governments should ensure "that news outlets, digital platforms, and political leaders provide the best available knowledge" and enshrine a new to not "be lied to by the powerful" in a world optimised for enragement: "one study showing that each additional negative word in a headline increased the clic…

"The indiscriminate proliferation of AI-generated content will not empower the underrepresented or democratise knowledge creation... [but] dilute and fragment the authenticity and reliability of information ... the discerning judgment of human curators is the only defence against misinformation and mediocrity".Why? Human curators :"…

When James West "joined a small wave of users granted early access" to the new Bing (I'll call and tag it "BingGPT"), powered by the same LLM behind ChatGPT, which is "a great party trick ... powerful work tool, capable of jumpstarting creativity, automating mundane tasks", he soon "noticed strange inconsist…

Real Facebook Oversight Board's 8-page "Q2 Facebook Harms Report ... deep-dive into Facebook’s impact on democracy, privacy, and human life this financial quarter" sets out:"serious problem with disinformation" which they actively choose to do nothing abouthow they let Trump's affiliate PACs to raise money on Faceboo…

Looking at disinformation "from a technological angle ignores that disinformation is an idea", so "we asked what the greatest thinkers in the history of philosophy would make of disinformation".Starting with Plato, writing when Athens was socially and politically turbulent, caused by a new invention: writing and reading. Plato…

A "perverse downstream consequence for debunking... being corrected by another user for posting false political news increases subsequent sharing of low quality, partisan, and toxic content".Looks like evidence for the backfire effect: "Direct correction ... backfires by making people feel defensive or focusing their attention on so…

"social and psychological forces that make people prone to sharing and believing misinformation ... are on the rise" - not so much created by bad actors, but exploited by them. Why?“cognitive and memory limitations, directional motivations to defend or support some group identity or existing belief, and messages from other people and pol…

"GPT-3 ... generates tweets, pens poetry, summarizes emails, answers trivia questions, translates languages and even writes its own computer programs, all with very little prompting"It's surprised a lot of AI researchers, but also "often spews biased and toxic language" and isn't always convincing. It's a univers…

roughly 1 of every 1,000 pieces of content ... violates ... rules on hate speech... even with artificial intelligence and third-party moderators, the company was “deleting less than 5% of all of the hate speech posted to Facebook.”...civic integrity team... charged with protecting the democratic process and reducing misinformation ... recently dis…

What did I learn from Hubbing 50 resources and writing five editions on disinformation during the US elections?

DFRLab’s Foreign Interference Attribution Tracker (FIAT) database (see Interference2020.org) "captures allegations of foreign #us2020 interference... and assesses their credibility, bias, evidence, transparency, and impact".80 allegations were catalogued: a "sharp increase from 2016... vary widely in their evidence and objectivity, …

Fleeting Note (FN): the creation of Fediverse-based alternatives might suddenly become an urgency.

Trump's delegitimisation disinformation campaign bears fruit: while 60% of those polled by Economist/YouGov thought Biden legitimately won:86% of Trump voters said he hadn’t, with 80% saying he shouldn’t concede, and most thinking the courts will overturn the resultsover 75% think there was sufficient fraud to influence the outcome - although…

When changing their Newsfeed algorithm, Facebook's tests found "a dramatic impact on the reach of right-wing “junk sites,”... [by] January 2018 a second iteration dialed up the harm to progressive-leaning news organizations instead... Mother Jones was singled out as one that would suffer... Daily Wire was identified as one that would be…

"Facebook quietly stopped recommending that people join online groups dealing with political or social issues", and will "assess when to lift them afterwards, but they are temporary." This part of the Facebook's AI plays a key role in pushing "people down a path of radicalization... groups reinforce like-minded views …

An interesting corrective: here's the Tech columnist for the New York Times, claiming Facebook is flooded with right-wing misinformation, not all labelled as such - and "it's not clear that labels are doing much". Most people - including, usually, me - will read that tweet, have their beliefs confirmed and move on. What's…

the Election Integrity Partnership and Disinfo Defense League ... two initiatives among hundreds ... reflect two mutually reinforcing sides of the same ecosystem... people devoting their livelihoods to combatting online misinformation and disinformation... These are their stories...EIP is uniquely close to the platforms, using Jira to communicate …

Several of the president’s tweets overlaid with disclaimers ... invoking Twitter's Civic Integrity Policy, which prevents users from manipulating elections. Trump’s tweets also effectively throttled by preventing other users from amplifying them. Facebook’s measures were less accusatory, attaching an informational label.Neither stopped Trump …

Every day, Times reporters will chronicle and debunk false and misleading information that is going viral online... Whole collection of individual stories, all worth a look. Some key points:Twitter flagged half of President Trump’s 14 posts on Thursday for including disputed or misleading information. But hundreds of other accounts - over 150 of w…

"Falun Gong-backed newspaper used aggressive Facebook tactics and misinformation to create an anti-China, pro-Trump media empire" - from a "low-budget newspaper ... handed out free on New York street corners" to "one of the country’s most powerful digital publishers... a leading purveyor of right-wing misinformation"…

Facebook threatens with legal action NY academics seeking to study Facebook's political ad targeting because their Ad Observer is a "scraping tool". The project collects data unavailable on Facebook's political ad archive: targeting data, geography, political contest, etc.., which they say now violates their rules established i…

Another analysis - with a lot of useful links - of the way social platforms managed the NYPost story, and what it tells us "about the status of social media companies’ reform efforts, and [how]... they are still failing to safeguard the democratic process?"First some facts:The NYPost article's "sourcing was so dubious that at l…

The platforms started doing what people concerned about disinformation have been asking for, and immediately came under fire from people concerned about censorship.When The New York Post published a story supposedly incriminating Hunter Biden "YouTube largely did nothing, Facebook deprioritized the Post story (slowing spread until 3rd party f…

There are two possible reasons why we are not talking as much about foreign interference. Both could be true. Only one is good news.

A few days after Facebook, Youtube bans "content that targets an individual or group with conspiracy theories that have been used to justify real-world violence... such as QAnon or Pizzagate", although they may except "news coverage on these issues or content discussing them without targeting individuals or protected groups".&q;…

"church officials don’t have a key to the box and that GOP officials picks up the ballots". Meanwhile, none of the ballot collection locations listed on the local Republican Party's website are official - instead they list their HQ, gun shops and other local businesses.They win either way: they can pass on Republican votes and suppr…

Most of the changes will start on Oct. 20, incl.:extra friction on retweeting, encouraging quote RTs adding the users' own thoughtspre-sharing warnings about flagged contentlabelling early election result claims disabling the algorithmic timelineAlso increasing factchecking and additional context on trending tweets.

Berkman Klein Center analyzed 55,000 news stories, five million tweets, and 75,000 public Facebook posts to discover that it's "Trump, the Republican National Committee, and Fox News — not Facebook spammers and Russian trolls — who are the primary drivers of misinformation around mail-in voting fraud", taking advantage of three stan…

WEF examines the impact deepfakes could have on US2020. Not a lot of the analysis is new, except perhaps:how deepfakes could be used subtly: making small changes to "how a candidate delivers an otherwise authentic speech ... [to call] character, fitness and mental health into question"some examples from Gabon, Malaysia & Belgium(?)Ca…

"it would be useful to see if an account is a repeat offender or has been repeatedly labeled".Research undermining the backfire effect is now leading Twitter to a "more obvious" label design, and are exploring virality prediction to help speed up labelling. Difficult to assess effectiveness - insufficient public data.