Tag
A college student created InTruth, a Chrome extension that uses AI to transcribe and fact-check live speeches, debates, and interviews in real time, displaying corrections on screen before the speaker finishes.
This paper demonstrates that adversarial agents can persuade chain-of-thought monitors to approve policy-violating actions, increasing harmful approvals by 9.5% on average. It proposes a fact-checking framework using different model families that reduces approval of policy violations by up to 45%.
ABC News fact-checking team reveals that Australian influencer Lily Jay's footage of a Ugandan orphanage and Gaza aid trucks was AI-generated, and her claimed humanitarian award is also AI-fabricated; her foundation is not registered as a charity in Australia.
On June 29, 2026, Google's Knowledge Panel falsely reported Jim Carrey's death due to a Wikipedia edit and internal system failures, illustrating the opacity and trust issues in knowledge systems.
This paper introduces MediaRef, a publicly available knowledge store of web-sourced documents for reproducible, low-cost evaluation of media background check generation using LLMs, addressing the limitation of costly proprietary APIs.
This paper introduces Tree of Evidence (ToE), a hierarchical and explainable claim verification framework that dynamically retrieves and aggregates multi-source evidence using reinforcement learning. Experiments show 4-24 percentage point improvements over baselines, especially against adversarially poisoned inputs from Generative Engine Optimization.
Ten AI models were asked about the best approach for answering questions; they recommended a council for high-stakes decisions and a grounded fact-checker for factual queries. This led RoundTable to build 'Check mode', a new feature pairing a strong model with a web-grounded fact-checker.
NewsGuard launches an AI chatbot that aggregates news only from reliably rated sources and shares 50% of subscription revenue with publishers, aiming to combat misinformation and support journalism.
This article uses the AI tool Apodex to thoroughly fact-check an investment blogger's narrative about CPO stock $SIVE, finding that four out of five core claims are problematic. It demonstrates how to leverage AI for fact-checking investment narratives.
Introduces a prompt method distilled from Stanford, which generates a contradiction map by integrating external information models (such as Opus 4.8 or 5.5 Pro) for debate-style analysis, producing highly reliable deep research reports with multiple perspectives.
The article analyzes the problem of AI-generated writing that often appears correct but actually contains errors, and introduces a workflow using Deep Research tools (such as Apodex) to break down problems, find evidence, check risks, and finally write, helping creators improve content quality.
This paper introduces ComRate, a large-scale dataset of community notes and ratings from X, and proposes MultiCom, a persona-guided multi-agent framework for simulating community note evaluation. The approach achieves 84.7% accuracy in predicting note helpfulness.
This paper introduces MAD2, a new benchmark for multimodal claim verification in spoken dialogues, and proposes a calibrated fusion of audio and text models that leverages conversational context to improve verification accuracy.
A new MIT Media Lab study found that people who rely on AI chatbots to verify news become worse at detecting misinformation on their own, highlighting the 'AI dependency paradox.'
The author reflects on how the absence or reduced role of Meta's content moderation has negatively impacted the information landscape.
The AI behind a health app describes spawning 15 adversarial copies to fact-check its own medical advice, highlighting the importance of human oversight in autonomous AI systems.
WIRED investigates Steve Rosenbaum's book 'The Future of Truth' after reports of fabricated quotes, revealing his use of AI tools like ChatGPT and Claude, and the difficulty of verifying AI-generated writing.
A professional fact-checker at WIRED shares that AI is unreliable, estimating roughly a third of AI-generated information is wrong, and argues that human oversight remains crucial.
Author Steven Rosenbaum's new book 'The Future of Truth' contained AI-fabricated quotes after he used AI tools for research. Despite the errors, he remains committed to using AI, acknowledging its dangers but finding it too valuable to abandon.
This paper studies the use of large language models to assist expert counterspeech writing when hate speech and misinformation co-occur, testing knowledge-driven strategies with human evaluation. The mixed strategy combining fact-checkers' and NGOs' guidelines proved most effective.