AI Bans in Client Contracts: What Practitioners Can and Cannot Do

·1 views

A growing number of financial professionals are using AI tools less frequently not due to capability issues, but because client contracts now include clauses restricting generative AI use. Legal experts note that such clauses typically fall into three distinct categories: prohibiting model training on client data, banning consumer-grade tools, or barring all third-party processing entirely. Commercial API accounts with data processing agreements often satisfy the first two restrictions, as major providers like OpenAI and Anthropic document that API-tier data is not used for model training by default. The broadest clause type, which prohibits any third-party processing, effectively rules out all hosted AI models and limits practitioners to local deployments. Professionals are advised to obtain the exact contract language and clarify ambiguities with clients rather than defaulting to the most restrictive interpretation.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

Developer builds interactive bilingual landing page inspired by Venezuelan Andean soup

A developer created PISCA, an interactive bilingual landing page dedicated to pisca andina, a family of breakfast soups from the Venezuelan Andes, as part of the Frontend Challenge: Comfort Food Edition. The page features a central visual concept where the soup acts as the only warm light source in a dark pre-dawn mountain landscape, with its steam doubling as Andean ridgelines. A range control titled 'Warm the morning' lets users transition through three states, progressively raising the sun and brightening the scene. The project was built using Vite and TypeScript with no UI framework, supports English and Spanish, and uses entirely original SVG and CSS artwork with no external raster images. Accessibility was treated as a core design principle, incorporating keyboard navigation, reduced-motion support, semantic landmarks, and bilingual accessible labels throughout.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Why reversing affiliate commissions on Stripe payment failures can backfire

Affiliate programs built on Stripe subscriptions face a subtle but costly flaw in how failed renewal payments are handled. The event 'invoice.payment_failed' fires once per retry attempt — up to eight times in two weeks — meaning a naive handler can claw back the same commission multiple times. Worse, many failures are temporary, so reversing a commission early can shortchange affiliates on payments that eventually succeed. A more reliable approach introduces a third commission state, such as 'pending_recovery', that holds the payout during the retry window rather than immediately reversing or paying it out. Terminal reversal should only trigger on confirmed uncollectible or voided invoices, though developers must note that 'invoice.marked_uncollectible' is a manual status and may never fire automatically.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Dev's game broke on GitHub clone because dist/ folder was blocked by .gitignore

A developer discovered their web game worked perfectly on Itch.io but failed completely when cloned from GitHub, showing a blank, unclickable canvas. The root cause was a .gitignore rule explicitly excluding the dist/ folder, meaning Git never tracked the production build files. Itch.io worked because builds were manually zipped and uploaded locally, bypassing Git entirely, while GitHub received a repo missing the critical script files. The fix involved skipping the dist/ output path and pointing HTML asset links directly to raw source files instead. After updating and pushing the corrected paths, the cloned repository loaded the game successfully with all event listeners intact.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

MemoFS Stores AI Agent Memory as Local Files to Fix Session Amnesia

AI agents routinely lose context between sessions because large language models rely on token-based working memory rather than persistent storage, causing them to forget past decisions and architectural rules. Expanding context windows to over one million tokens has not solved the problem, instead introducing issues like context compaction, attention degradation, and rising API costs. Vector databases, a common alternative, are difficult to inspect, lack version control integration, and perform poorly on exact-match queries involving file paths or symbol names. MemoFS addresses these limitations by storing all agent memory as plain Markdown and JSON files inside a project's local .memofs/ directory, making memory human-readable, grep-able, and version-controllable. The tool uses a hybrid BM25 and vector index alongside a knowledge graph to give agents durable, auditable, and branch-aware memory without relying on remote databases.

0 comments Read more at DEV Community