The SpicyRanks briefing

The news behind the conversation.

AI companions, adult platforms, and the technology behind them. The big updates, the hard questions, and the sources to back them up.

AI technology/ News

Inkling 2 Mini brings open weights and a looped design to local hardware

Thinking Machines' November 20 release offers downloadable weights and Tinker access. Its looped architecture reuses middle layers, while the model card specifies memory requirements and the announcement acknowledges weaknesses in multi-turn chat.

· 2 sources
AI technology/ News

Liquid AI adds audio and typed extraction to d1.2 decision models

Liquid AI's November 16 release extends its hosted decision models beyond selecting options. The practical question is whether their typed outputs and missing-value behavior fit a workflow—not whether a company benchmark makes every decision dependable.

· 2 sources
Guides & analysis/ Blog

When your AI reads a webpage, who gets to give it orders?

A document can contain instructions the user never authorized. OWASP's guidance explains why keeping external content separate, limiting tool powers and checking consequential actions are different protections. This guide offers reader questions, not a product security certification.

· 2 sources
AI technology/ News

Microsoft launches MAI Code 2 and Flash, with wider access still promised

Microsoft's November 17 coding-model announcement separates Copilot availability, a limited Foundry preview and future distribution. Its benchmark cost claims use specific configurations and promotional prices, rather than establishing universal savings.

· 1 source
Guides & analysis/ Blog

Choose the image edit before choosing the AI control

Start with what must change and what must remain. Current Midjourney and Black Forest Labs documentation shows why targeted edits, wider canvases and whole-image transformations deserve different workflows. This is a documented guide, not a hands-on comparison.

· 3 sources
Guides & analysis/ Blog

What your AI subscription’s credit count leaves out

A credit balance does not answer every buying question. Separate what consumes credits, when allowances reset, which features your plan unlocks and whether spending can continue automatically. ElevenLabs’ published rules show why those distinctions matter.

· 3 sources
Guides & analysis/ Blog

A confident AI answer still needs an evidence check

Treat confidence as a cue to inspect evidence, not a substitute for it. A useful verification habit checks what the source actually says, which claims it supports and where the answer goes beyond it.

· 2 sources
AI technology/ News

Google launches a Gemini API for bounded decisions

Google's November 11 launch targets choices such as routing and rubric scoring. Its practical distinction is a structured decision interface, not another open-ended chat box.

· 1 source
Guides & analysis/ Blog

Local or cloud AI? Follow the data, then count the tradeoffs

Local and cloud describe where work happens, not a complete privacy or quality verdict. Start with the route your prompts take, then consider storage, hardware and the features you actually need.

· 3 sources