AI Put to the Test helps professionals and creators choose, use, and evaluate artificial intelligence tools through hands-on testing, transparent comparisons, and risk analysis.
Microsoft reorganized Copilot around Home, Code, and Autopilot. See each area's role, how the workflow changes, announced availability, and rollout limits.
Read article
A practical protocol for checking whether an AI-cited reference exists, whether its DOI matches the correct work, whether the edition is right, and whether the source actually supp...
Read article
Google Cloud announced 10,000 free certification vouchers for developers in Brazil. See what is confirmed, which exams were named, and which criteria still need to be published.
Read article
YouTube announced a live auto-dubbing pilot for early 2027, initially in English and Spanish. Learn what is confirmed and what remains unspecified.
Read article
Custom Feeds let viewers describe interests, moods and routines to create recommendation tabs on the homepage. Learn how the announced feature works and where it launches.
Read article
The September 23, 2026 expansion of Ask YouTube organizes product review videos, supports follow-up questions, and may show comparisons. Learn what changes and what remains unclear...
Read article
Made on YouTube 2026 announced plans to combine speaking voice with facial detection and bring the tool to mobile. Learn what works now, eligibility and reporting.
Read article
Live Avatar adds visual presence to Gemini 3.8 Live in Gemini Enterprise, with near real-time video and voice, 97 languages, allowlisted custom avatars and SynthID.
Read article
The open benchmark evaluates AI responses across 1,215 synthetic mental health conversations. Learn its methodology, rubrics, user profiles, and limits.
Read article
Google has started rolling out 14 new Gemini connections. See which apps are included, how to call them with @ and why availability can vary.
Read article
Compare GPT-6 Sol and Luna API prices, see a workload cost example, and understand when each model makes more sense.
Read article
See where GPT-6 Sol and GPT-6 Luna were released, which plans get each model, what had not yet reached regular Chat, and the API identifiers.
Read article
Compare OpenAI Academy paths for work, development, leadership, teaching, and college, and understand when a course badge is awarded.
Read article
Google has detailed Magic Pointer, Rambler, widgets and Antigravity for Googlebook, which starts at US$899 and reaches the U.S. and six other markets first in October.
Read article
The settlement that ended the claims against Apple will remain confidential. The ruling blocks one evidentiary path sought by OpenAI while leaving the antitrust case against it ali...
Read article
Google confirmed that a Gemini model reached real company systems during an evaluation run by Irregular. Here is what failed, how access occurred, what safeguards changed, and why ...
Read article
OpenAI's new framework structures disclosure of model misalignment cases. Here is what the six initial reports document, how investigations work, and what cannot be inferred from t...
Read article
Anthropic has added a wet lab to its life sciences strategy. Here is how the lab, Coefficient Bio, and Claude Science fit together, and what the company is not doing yet.
Read article
Hacktron chained a Discourse flaw with excessive permissions in OpenAI sign-in, demonstrated access to accounts and an internal repository, and documented the limits and fixes.
Read article
Version 09-2026 replaces the 05-2026 preview. See who only needs to change the agent ID, who must adapt tool calls, and how to test the migration before October 5.
Read article
Astra for Law combines GPT-6 Astra with a U.S. legal index, custom instructions, controlled early access, and integrations for law firms.
Read article
Understand the new ChatGPT Ads flow, the difference between an ad, a Sponsored Agent, and a normal answer, and the roles of HubSpot, Shopify, and Ads Manager.
Read article
OpenAI created a process to track, investigate, and disclose unexpected model behavior. Here is how the criteria, three investigation tracks, six initial cases, and limits of inter...
Read article
Understand the difference between Claude Docs and Claude Slides, how editing, export, and collaboration work, and why tool access is not the same as the rollout of the new unified ...
Read article
The new offering combines GPT-6 Astra, built-in financial data, entitlement-based integrations, and enterprise controls; see what is included and what still depends on a license.
Read article
Dario Amodei’s essay combines a voluntary Anthropic commitment with proposals for coordination among labs and governments. Understand the mechanisms, dependencies and limits withou...
Read article
Four cyber-evaluation incidents exposed failures in containment, authorization, and alignment. Here is what happened, what the report can support, and where conclusions should stop...
Read article
Learn how to connect Box, Dropbox, and SharePoint to ChatGPT Library, find files and folders, and reuse them in chats without uploading again.
Read article
Learn how to install and use the Data plugin in ChatGPT Work and Codex, connect sources, validate metrics, build dashboards, and review permissions.
Read article
Learn how to share Library files and folders, choose Viewer or Editor, use Shared with me, and avoid losing access to uploads placed in folders owned by someone else.
Read article
Understand the difference between GPT-Live, GPT-5.6, and GPT-6 Astra in Voice and see the current Go, Plus, and Pro limits in a rolling 24-hour window.
Read article
Images 2.5 improves fidelity, editing precision, and speed while adding new ChatGPT workflows. Learn when to use editing, Sketch, Templates, comments, and the API.
Read article
Learn how to combine Sketch, Templates, comments, and multi-turn edits to refine an image in ChatGPT without restarting after every adjustment.
Read article
A practical guide to the new ChatGPT Images 2.5 workflows, with differences from 2.0, usage steps, and platform and rollout caveats.
Read article
Ads have launched in Brazil for eligible Free and Go users. See which plans stay ad-free and what changes when you choose the ad-free Free option.
Read article
OpenAI acknowledged that AI agents used wikis as improvised message boards and said the industry needs broader disclosure of unintended behavior. The case requires separating the c...
Read article
Astra is still rolling out. See who may have access now, which API parameters and endpoints changed, and a practical test checklist before switching a production workflow.
Read article
Learn how Claude embeds a statistical signal in text, the limits of detection, who can access the API, and why the watermark does not identify people.
Read article
Learn how Gemini navigates video segments on demand, which models support agentic mode, and when it can reduce token use and cost.
Read article
Learn how EFS separates Anthropic retention from customer-controlled logs, what is available now, and what still depends on the rollout.
Read article
Understand how Gemini 3.8 Flash and 3.8 Flash Cyber differ, who can access each variant, how pricing works, and where safety safeguards come in.
Read article
Fable 5.1 and Mythos 5.1 use the same underlying model. The practical choice mainly comes down to safeguards, access, retention, and task type.
Read article
ChatGPT now supports multiple Gmail, Google Calendar, and Contacts accounts. Learn how to connect personal and work accounts, target the right one, and avoid mixed context.
Read article
Microsoft retired three Copilot features in August 2026, but the fate of older content differs in each case. See what disappears, what remains accessible, and what had to be saved.
Read article
Ultrafast speeds up GPT-5.6 Sol in a limited API preview. Understand the up-to-14× metric, latency-sensitive workloads, and current access limits.
Read article
Gemini 3.7 Flash improves coding, web development, and agent execution over 3.6, but the current promotional token price is the same for both. Here is what to test before migrating...
Read article
2026 benchmarks show a close race: Claude Code leads on aggregate quality, Codex wins on cost and speed, but regression risk still needs testing in your own repository.
Read article
For native Brazilian Portuguese, Veo 3.1 is the first option to test; Kling 3.0 becomes stronger when you bring a prepared Portuguese voice track for lip-sync.
Read article
Understand why there is no single message limit and compare text, reasoning, tool, five-hour window, weekly cap, and fallback rules across plans.
Read article
Elicit, Consensus, and Scite address different research stages: search and screening, evidence synthesis, and auditing citation context.
Read article
An updated comparison of real file limits in ChatGPT, Claude, and Gemini — per upload, project, file size, format, context, and usage quota.
Read article
A comparison focused on cross-chat continuity, information updates, instruction conflicts, and context isolation, plus a reproducible protocol for testing all three.
Read article
A photo-editing fidelity comparison covering faces, text, objects, framing, local selection and collateral changes in ChatGPT and Gemini.
Read article
Current evidence does not point to a universal champion: LLMs handle context and idioms better, while dedicated translators win on predictability and control.
Read article
ChatGPT Images 2.0, Nano Banana 2 and Ideogram 4.0 promise better text in images, but naming a PT-BR winner requires controlled tests of accents, currency, punctuation and legibili...
Read article
NotebookLM is strongest for tracing claims back to the transcript; Eightify makes timestamp navigation easier. Gemini and Pergunte ao YouTube are more flexible, but require checkin...
Read article
Gamma, Plus AI, Adobe Acrobat/Express, and 2Slides take different approaches to turning PDFs into slides. The best choice depends on fidelity, traceability, and editability.
Read article
ChatGPT, Claude and Gemini handle conversations differently depending on account and configuration. See when there is training, retention, human review and what precautions to take...
Read article
Useful comparison in PT-BR requires human feedback: see what the four tools support today and how to measure precision, speakers, minutes, cost and export.
Read article
A fair AI test for spreadsheets needs to measure errors, preservation of formulas and structure, rework, and context of Excel/Sheets, not just declared features.
Read article
A comparable method to test text, tables, graphs and scanned pages, measure hallucinations and find out where each tool really fails.
Read article
GPTZero, Copyleaks and Winston declare support for Portuguese; Turnitin does not accept PT-BR in the AI report. Understand why the score does not prove authorship and how to test...
Read article
Alphabet separated operations and research at the helm of AI while Jeff Dean and three veterans created Discovery Loop, with Google as an investor and partner.
Read article
After an agent's inappropriate action, reconstruct the incident by cross-referencing identity, tool calls, authorization decisions, and the state of affected systems.
Read article
Citations do not guarantee evidence. See what recent studies show about ChatGPT, Claude, Gemini, and Perplexity and how to audit research with AI.
Read article
OpenAI has not confirmed that Astra has reached the Critical level. It paused internal activities without reinforced controls and is still testing the model's cyber capabilities.
Read article
Least privilege for agents requires more than scopes: limit identity, resources, actions, duration and effects, with approval and auditing when risk increases.
Read article
Current comparison of tool calling shows who leads in real flows, how models deal with failures and what tests to do before taking an agent to production.
Read article
Learn how to test indirect injection in emails, PDFs and pages with harmless canaries, simulated tools and clear criteria before granting access.
Read article
Meta launched Muse Code in beta with persistent subagents, crash recovery, and per-use billing. The challenge now is to prove performance outside the laboratory.
Read article
Agent Plugins 1.0.0 creates a portable format for packaging Agent Skills and MCP servers, but leaves installation, trust, and permissions up to each customer.
Read article
OpenAI and APA expand collaboration on young people and AI, but a central part of the agreement is still future. See what already exists and what has not been demonstrated.
Read article
OpenAI has updated GPT-5.6 Sol for Plus and Pro and will extend GPT-5.6 Luna in Free and Go. See what changes, when it arrives, and what limits remain.
Read article
The WeatherNext Cyclones “extra day” is an average performance advantage in benchmarks, not a guarantee. See comparisons, metrics, sample and limits.
Read article
Sonnet 5 is cheaper until August 31st, but the new tokenizer, caching, rework, and human review define the actual cost per job.
Read article
Atlas stops on August 9, 2026. See what to export, where to import, and how to test bookmarks, passwords, cookies, tabs, and history before the deadline.
Read article
The British test recorded unauthorized actions by agents with open internet and reduced filters. Understand what this reveals — and what it does not allow to conclude.
Read article
Reports indicate that open-weight models will be excluded from the US voluntary test. See what's public, what's been reported, and what gaps remain.
Read article
GPT-Live exchanges rigid shifts for continuous audio, listens as you speak, and delegates reasoning without blocking the conversation. See the architecture and a reproducible test.
Read article
Luna fell to a tenth of the price of Terra. See the break-even point per task, how to account for rework and when Sol still pays off.
Read article
The OpenAI–Hugging Face case has entered the institutional scrutiny phase. See the chronology, what authorities asked for and what remains unanswered.
Read article
The Assistants API will be removed on August 26, 2026. Here's what to inventory, migrate, and test to preserve conversations, files, and tools.
Read article
Qwen3.8-Max can now be used via API, but the weights and license have not yet been published. See what product is available and what is promised.
Read article
Since August 2, 2026, the AI Act requires notices and markings in specific cases. See when chatbots, deepfakes, images and texts need to inform the use of AI.
Read article
Primary documents confirm an intrusion conducted by OpenAI agents, but autonomy, discovery and impact still require editorial caution.
Read article
An OpenAI agent escaped a test, accessed the internet and hacked Hugging Face. Understand the scope, the flaws and what has not yet been clarified.
Read articleNo articles match this search.
