This website uses cookies
Read our Privacy policy and Terms of use for more information.
We believe in the power of digital tools and digital media to improve access to information and improve democratic discourse. We recognize that economic and political incentive structures often lead to the proliferation of falsehoods, platform manipulation, scams, and other harms.
We believe in journalism as an accountability mechanism that furthers transparency and democratic discourse.
We focus on exposing digital deception in its many forms — scams, search engine and social media manipulation, disinformation, trolling, mobile app abuse, spyware, AI slop and more — because we believe it causes real harm and undermines the ability of citizens to understand and engage with the world around them.
We prioritize investigating digital platforms because of their reach. We aim to choose topics based on prevalence and harm.
We agree with the perspective that “freedom of speech does not equal freedom of reach.” We think platforms can and should impose rules that discourage abusive content and encourage healthy online ecosystems — or face consequences.
We generally believe that people who work for big platforms, especially in digital safety roles, care about the issues we cover and do their best to reduce harm. We also recognize they face resource constraints and internal incentive structures that don’t always align with information integrity.
We recognize the double-edged nature of open source intelligence tools: they can be used to responsibly gather information and can also be used for surveillance. We do our best to flag this in our coverage and to monitor for known abuse.
We believe socio-technical challenges require socio-technical solutions. Put differently: we don’t think that product fixes alone can solve societal problems.
We are primarily a reader-supported publication. In our first year of operation, subscriptions represented more than 90% of Indicator’s total revenue. However, we do run ads in our Briefing newsletter and Show & Tell podcast. Ads are clearly marked as sponsored content. We do not accept financial support from companies that we cover (as Indicator or as individuals)
We strive to be fair, accurate, and nonpartisan in our reporting. We do not belong to or donate to political parties. When we express an opinion in our work, we do so based on reporting and domain expertise.
Alexios used to work for Google and ran the International Fact-Checking Network. He holds no shares in the former. He discloses his role in the latter any time he writes about it.
We believe journalism can and should be used by policymakers, but do not lobby or engage privately with political decision-makers to promote specific regulations.
We are curious about digital tools and how they can be used in reporting, investigations, and trust and safety work. Part of Indicator’s mission is to keep on top of the tools and techniques that can help expose manipulation. For this reason, we actively use and test a wide range of tools, including those that use artificial intelligence. At the same time, we cover AI harms extensively and know the many weaknesses and flaws of its applications.
In order to use AI responsibly in the service of accountability journalism, we abide by the following principles:
Never delegate thinking: remain both curious and skeptical
We do not use AI to replace reporting. It doesn’t supplant our thinking or analysis. We align with the Associated Press standard that “any output from a generative AI tool should be treated as unvetted source material.”
Disclose substantial contributions proportionately
We describe how we use AI in our reporting, based on the weight of the technological contribution. Using AI to analyze a dataset will result in a more prominent disclosure than using it for scraping, for example. Our disclosures are typically made in the Indicator Info Box at the end of articles.
Only use AI for tasks you can QA
We build quality assurance into our AI-assisted work. We always verify all or a sufficient sample (typically 10%) of the data we analyzed with an AI tool, and review the exact steps taken.
Our use cases of AI include:
Assistance with scraping, gathering, organizing, categorizing data, and expanding on information that we’ve gathered. For example, we might use Claude in the browser to scrape view counts and comments from hundreds of TikTok scam videos we are writing about.
Assistance with reviewing newsletter and Google Alerts content to prioritize information and identify links and information of interest.
Generation of interview and workshop transcripts. For example, we use Google’s AI tools to generate transcripts and notes from our monthly workshops.
Drafting social media posts and other marketing material. Such content is never published without human review and editing.
Generation of dashboards, graphics, and charts, interactive or static. This always involves human review and vetting. We also check the resulting visualizations to ensure the data is represented accurately. One example is our AI Community Notes dashboard.
Generation of illustrations where we might otherwise have used stock imagery or screenshots. We label the image through an overlay or in credits
Using AI in our operations to access and analyze data about our audience, revenue, and subscriptions through Stripe and Beehiiv.
We use AI in our operations to access and analyze data about our audience, revenue, and subscription trends through Stripe and Beehiiv.
Tracking our impact. For example we automatically extract information about platform takedowns from our articles and cite the data on our About page. We manually review all outputs before publication.
Building tools for our members. We will test for a range of possible harms through a battery of representative and adversarial prompts that we make available to subscribers upon request.
We correct our errors publicly. We encourage people to contact us to request or suggest corrections at [email protected] and we flag when such an email leads to a correction.
We add corrections to the top of articles and also collect them on this page.
We aim to clearly state what was incorrect and to provide the correct information. We also aim to concisely explain why or how the error occurred.
We add a Correction when we correct a factual error of any nature. We add an Update when we insert information or improve the clarity of a passage in a way that does not materially change a statement or factual claim.
Corrected articles
Apr 14 2026: The Briefing was updated to better represent Spotify’s AI policy.
Nov 7 2025: The Briefing was corrected to accurately describe arXiv’s updated policy on paper submissions.