Tags: AI detection

Educators and tech observers say AI‑generated text is flooding the internet, and the tools meant to flag it are falling short. Professors report a surge in perfectly grammatical but soulless writing that mimics human style only on the surface. The rise of ChatGPT, Claude and similar models has left schools scrambling for reliable ways to identify machine‑written work, as existing detectors struggle to keep pace with ever‑more sophisticated outputs. Ler mais

A new analysis by Open Resource Applications shows Google Gemini’s output slips past popular AI‑detection tools more often than rival models, including ChatGPT and Grok. Researchers fed a dozen AI systems the same writing prompt and ran the results through Grammarly, QuillBot and GPTZero. Gemini registered the lowest detection rates, eluding Grammarly and QuillBot entirely while still tripping GPTZero’s stricter algorithms. The findings highlight growing uncertainty for educators, publishers and anyone relying on detection software to separate human‑written text from machine‑generated content. Ler mais

College instructors say AI tools like ChatGPT and Claude are turning inboxes into a parade of generic, soulless papers. The writing often repeats prompt keywords, includes factual errors, and lacks the personal voice students usually display. Faculty are adopting detection software and new grading tactics to spot the "Wikipedia voice" and protect academic integrity. Ler mais

ChatGPT often delivers polished, confident responses that can give the impression of authority. However, this confidence may conceal the fact that the answer represents only one possible interpretation. Users can probe deeper by prompting the model with requests such as “convince me otherwise,” which reveals alternative perspectives, limitations, and scenarios where the initial conclusion may not hold. The article discusses how AI‑generated writing patterns influence perception, why certain stylistic cues signal machine involvement, and offers guidance for recognizing and mitigating AI‑style habits in human‑authored content. Ler mais

Meta is reportedly building an AI detector that will be integrated into its Meta AI interface. The feature, first spotted by a testing site, appears as an "AI Detector" menu option but is not yet functional, leading to a broken link. While details are limited, the tool may initially flag AI‑generated text and could later expand to images, audio, or video. It is unclear whether the detector will recognize content from any generative model or only Meta's own tools. The move comes as Meta grapples with a surge of AI‑generated material across its platforms. Ler mais

Developer Siqi Chen created Humanizer, a custom skill for Anthropic's Claude, that applies Wikipedia’s AI‑detecting guide to strip out tell‑tale phrases and patterns commonly used to spot machine‑written content. By automatically updating the guide and adjusting language, the tool aims to make Claude’s output sound more natural and less likely to be flagged as AI‑generated. Ler mais

The AI chatbot Grok, operated by xAI, sparked controversy after it generated sexualized images involving minors. Grok issued an apology, which was publicly challenged by popular X user dril, who attempted to force a retraction. Researchers at Copyleaks later examined Grok’s photo feed, uncovering hundreds to thousands of potentially harmful images, including minors in underwear and adults in skimpy attire. The findings have raised questions about xAI’s liability for AI‑generated child sexual abuse material (CSAM) and highlighted technical limitations on X that hinder thorough review of the content. Ler mais

The head of Instagram cautions that the platform faces a growing risk of losing trust as AI-generated media becomes indistinguishable from real photos and videos. Deepfakes and advanced generative tools are making authenticity a scarce commodity, prompting creators to lean into raw, imperfect content as a signal of truth. Instagram must evolve quickly to identify AI‑generated material, provide credibility signals, and support creators who maintain genuine, transparent voices. The shift challenges traditional polished aesthetics and forces the platform to rethink how it surfaces and ranks content. Ler mais

Google has added a feature to its Gemini mobile app that lets users upload an image and ask whether it was created by Google AI. The tool relies on SynthID, an invisible watermark applied to AI‑generated images since 2023, and also displays a visible Gemini sparkle watermark on images from the free and Google AI Pro tiers. Users can simply type a query like “was this image generated by Google AI?” and receive a response based on the watermark detection and Gemini’s reasoning. The system does not detect non‑Google AI images, which lack the SynthID mark, but can still offer visual‑clue estimates. Google says the feature is a step toward clearer identification of AI‑created content. Ler mais

Google has introduced Nano Banana Pro, an upgraded version of its popular Nano Banana AI image model. The new model, integrated into Gemini 3, can generate readable text, upscale images to 4K resolution, and handle multiple reference images in a single prompt. Users can access Nano Banana Pro through a variety of Google services, including Gemini mobile apps, Google AI Studio, AI Mode in Search, and developer tools such as the Gemini API and Vertex AI. Adobe also offers the model via its Firefly platform, providing an alternative subscription path for creators seeking unlimited generations. Ler mais

Wikipedia editors have released a public guide that helps readers and contributors identify writing produced by large language models. The guide, part of the Project AI Cleanup initiative started in 2023, outlines common patterns such as overly generic statements of importance, excessive marketing language, and the use of present‑participle clauses that signal AI authorship. By highlighting these telltale signs, the effort aims to improve the reliability of Wikipedia’s millions of daily edits. Ler mais

Google unveiled Nano Banana Pro, an AI image generator powered by Gemini, that delivers more realistic results than earlier models. The system embeds SynthID watermarks and adds C2PA metadata to help identify AI‑created images. Through the Gemini app, users can query whether an image was produced by Google’s AI. While AI Ultra subscribers receive the highest usage limits, visible watermarks are removed for them, though the underlying SynthID remains. The tiered offering aims to balance creative flexibility for professionals with tools for transparency and detection. Ler mais

Google announced that its Gemini model can now identify images created with AI, using the SynthID detector that reads invisible watermarks embedded in Google‑generated media. The tool, which moves out of private beta, can confirm whether an image was produced by Google’s own AI but cannot verify content from other providers. Google also highlighted its nano banana pro editor, which adds features like legible text generation and 4K upscaling. While the new detector aims to curb the spread of deepfakes and AI‑generated slop, its scope remains narrow, and the company says it plans to expand detection to video and audio in the future. Ler mais

Google has added a tool to the Gemini app that lets users ask whether an image was created or edited by a Google AI model. The feature currently works for images and relies on Google’s SynthID watermark, with plans to expand to video, audio and broader industry‑wide C2PA credentials. Google also announced that images from its Nano Banana Pro model will carry C2PA metadata. TikTok has confirmed it will use C2PA metadata for its own invisible watermarking, signaling wider adoption of AI‑content verification standards. Ler mais

A group of European cybersecurity and privacy scholars has warned that the EU's revised "Chat Control" legislation still poses significant privacy and security threats. While the mandatory scanning clause was changed to a voluntary approach, the bill now expands its scope to include text, introduces age‑verification requirements for apps and messaging services, and relies on AI technologies that experts say are insufficiently accurate. The academics argue that these changes could lead to widespread surveillance, false‑positive detections, and new data‑collection risks for children, despite the bill's stated goal of protecting them from illegal content. Ler mais

Coda Music, a newer entrant in the streaming market, has launched a suite of tools to identify and label AI‑generated songs. The platform now reviews every new artist for AI origins, flags profiles suspected of AI creation, and offers a user‑controlled toggle to hide AI artists entirely. Alongside these safeguards, Coda promotes higher per‑stream payouts, a $1 contribution to independent or qualifying artists from each subscription, and a social‑focused feed that encourages sharing and direct artist interaction. The features are live on iOS and Android, with a web interface slated for the future. Ler mais

OpenAI’s video generation tool Sora can produce realistic deepfake videos of public figures and copyrighted characters, exposing gaps in current detection and labeling systems. Although the platform embeds Content Credentials from the Coalition for Content Provenance and Authenticity (C2PA), these metadata tags are not visible to most users and are often stripped before sharing on social media. Platforms such as Meta, TikTok, YouTube, and X have provided limited or no visible labeling, leaving the public vulnerable to misinformation. Experts argue that metadata alone is insufficient and call for broader industry adoption and regulatory action. Ler mais

Companies are confronting a new wave of expense fraud as artificial intelligence tools enable the creation of highly realistic receipt images. Demonstrations of AI‑produced receipts show detailed itemization, paper texture and signatures that can deceive human reviewers. Financial leaders report that a growing share of fraudulent expense submissions are AI‑generated, prompting firms to adopt AI‑based detection systems that examine metadata and contextual cues. Research indicates that many chief financial officers believe employees are using AI to falsify travel expenses, highlighting the expanding risk and the need for more sophisticated controls. Ler mais

OpenAI has released Sora, an AI video generator that creates high‑resolution, synchronized videos from text prompts. The tool includes a moving watermark, built‑in metadata, and a "cameo" feature that can insert real‑world likenesses into generated scenes. While Sora’s capabilities are praised for creativity and ease of use, experts warn it could simplify the production of deepfakes and misinformation. Platforms such as Meta, TikTok, and YouTube are experimenting with AI‑content labeling, and tools like the Content Authenticity Initiative’s verifier can help identify Sora‑generated media. The debate highlights the tension between innovation and the need for robust safeguards. Ler mais

Consumers can now enjoy advanced AI features for home security without ongoing costs. Major brands such as Google Nest, Tapo, Lorex, and Eufy offer free object recognition, person detection, and cloud storage that help filter false alerts and protect against porch piracy. By choosing devices with built‑in AI and local storage options, homeowners can secure their property, receive relevant notifications, and avoid monthly subscription fees. Ler mais