Context compaction drops 83% of user rules

Long chats get summarized so the model can keep going. The summary keeps the job. It throws away how you asked to be treated.

Penn State researchers find that when AI systems compress a long conversation to free context, they retain only 17 percent of the session rules users actually set — things like "don't send email without my approval" or "confirm with me before you change anything." Zhiqi Wang, Yichi Zhang, Dongwon Lee, and Yuchen Yang call these session constraints: instructions that bind how the model behaves for the rest of the chat, not what the job is. Compactors are built for task continuity. They keep the goal, the current state, and the next step. A one-line "ask first" rule is not the task, so it dies in the summary. After that, the agent will send the email.

The team built CompInt, a 750-instance suite that injects those constraints into three long-context settings — multi-turn chat, agent trajectories, and long-horizon research — then checks whether the compacted memory still contains them. Most compactors they tested, from a keep-the-last-five-turns baseline to LLMLingua-2 and prompt-based summarizers on open models, do worse than leaving the full context alone. With the uncompressed conversation still in the window, models follow the rule 59 to 71 percent of the time, depending on the setting. After compaction, compliance often sits only a few points above the no-rule baseline. A prompt written specifically to preserve user constraints still keeps retention below 40 percent on the chat dataset. GPT-5.4-mini is the one model that sometimes beats the uncompressed baseline — and even that result is brittle, swinging from 98 percent retention on research traces to under 7 percent on WildChat with the same prompt family.

The useful finding is that this is not a prompt problem. The authors attach a small add-on, built on Qwen3.5-9B, that reads each user turn, lifts out session constraints into a separate list, and staples that list back onto the summary. No training, no change to the main model or the compactor. Retention jumps to 95.6 percent on agent traces, 95.1 percent on research tasks, and 90.3 percent on multi-turn chat. The paper and the evaluator are public.

Labs sold context windows as memory. Compaction is how those windows actually work once a session gets real, and the summarizer is optimized to keep the project moving, not to remember that you said no. "Don't delete until I confirm" is exactly the kind of instruction that is not the task and is the whole point. If your product auto-compacts a long thread, you already have this hole. The fix is a separate register for user rules, not a nicer summary prompt.

What to watch: whether the major agent runtimes start treating session constraints as first-class state, the way they already treat the system prompt.

If your agent compacted a long chat this week, did it still honor the one rule you actually cared about? Tell us in the comments.

Sources: CompInt (GitHub)

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings

Read more

AI 101 — How do AI video generators work?

Houmo moves compute inside memory — and A20 enters production

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Brain organoids classified Braille at 83% — in a trio, not alone