memory
Stanford researchers compress a memory readout into 16 vectors News
Memorilla reads a persistent embedding store and supplies a question-specific 16-vector summary to a language model.
Google explains an offline sleep cycle for continual-learning models News
Google Research’s Sleep framework uses offline weight updates to consolidate temporary learning and train on selected synthetic experience.
ChatGPT locked chats hide conversations, but PIN reset deletes them News
OpenAI is rolling out PIN-protected locked chats, with permanent deletion of protected content when a forgotten PIN is reset.
A ChatGPT memory audit found sensitive context, not proof of a profile on every user News
An ACM study found psychological categories in 52% of 2,050 donated ChatGPT memory entries, with limits that prevent claims about every user or the current system.
Muse’s memory report raises profiling questions, while Meta describes separate user environments News
Reporting on Muse’s relationship files raises privacy questions, while Meta’s public disclosure describes per-user environments and sanitized product learning.
High-bandwidth memory: why AI chips are built around stacks of DRAM Lesson
High-bandwidth memory (HBM) stacks DRAM dies on top of each other beside the processor, giving AI accelerators the enormous data rates that generating text demands; its capacity and bandwidth now set how many models and users a data center can serve.
Epoch AI: chips shipped through 2027 could run 30-170 million frontier agents at once, or billions of cheaper ones News
Epoch AI estimates that the high-bandwidth memory shipping from 2025 through 2027 could support roughly 30 to 170 million always-on frontier-model agents, or about 1.9 billion running an efficient open model, if demand materializes.
Micron forecasts tighter memory and storage supply through 2028 News
Micron expects memory and storage supply-demand conditions to tighten in 2027 and 2028 despite expanding factory investment.
Google presents a four-part orchestration stack for coherent long-form video News
Google has grouped planning, persistent visual memory, segment repair and visual critique into a long-video production stack built around Gemini and Veo, with uneven evidence and release status across its four components.
Gradient accumulation: training with a large batch when it will not fit Lesson
Gradient accumulation adds gradients from several small forward-and-backward passes before updating parameters, letting a training job approximate a larger batch without storing that larger batch in memory at once.
Continual-learning preprint improves long-horizon memorization, not general learning News
A new preprint reports that combining replay, function and weight anchors with merged LoRA raises final retention in a narrow sequential Q&A test from 1.2% to 34.9%, while general abilities still deteriorate.
Gradient checkpointing: throwing work away so training fits in memory Lesson
Gradient checkpointing cuts the memory a neural network needs during training by deliberately discarding most intermediate results and recomputing them later, trading roughly 30% extra compute for a memory footprint that drops from linear in network depth to the square root of it.
DRAM contract prices nearly doubled in a single quarter News
Conventional memory contract prices rose roughly 93% to 98% quarter over quarter in early 2026 and are forecast to climb another 58% to 63%, as suppliers divert capacity to AI servers -- repricing the exact component local AI depends on.
Seven senators demand Apple reject Chinese memory chips as AI demand drains global supply News
A bipartisan Senate letter urges Apple to commit that no memory from Chinese suppliers CXMT or YMTC will appear in any Apple product worldwide, noting that CXMT turned profitable only after the AI-driven global memory shortage took hold.
A world model that keeps the map outside the model News
Evoke stores scene geometry in an external memory bank indexed by camera position and retrieves only what the current view needs, letting an interactive world model run open-ended sessions without its context growing.
Compressed memory stretched a 7,000-token model to 1.75 million News
A framework that chunks long documents, compresses each chunk into memory blocks and gates which blocks the model reads extrapolated from 7,000 tokens of training context to 1.75 million at inference, with half the peak GPU memory of a leading baseline.
China's biggest memory maker is booked through 2027 News
ChangXin Memory Technologies has reportedly sold out its DRAM output through the end of 2027 as PC brands rushed to secure supply, and consumer memory prices have stayed near their highs since.
High Bandwidth Flash Became a Spec Today, Not a Product You Can Buy News
SK hynix and Sandisk published the first standard for High Bandwidth Flash at FMS 2026, defining a NAND memory tier of up to 512GB per stack with a top bandwidth grade near three terabytes a second - with no named accelerator, price or availability date.
Offloading and streaming: running a model bigger than your memory Lesson
Offloading and streaming let a machine run a model far larger than its memory by keeping only the parts needed right now in fast memory and fetching the rest from system RAM or disk on demand, trading speed for capacity.
An $8 Microcontroller Now Runs a 28.9-Million-Parameter Language Model News
A developer fit a 28.9-million-parameter model onto an ESP32-S3 with 512KB of fast memory by keeping most of the model in flash storage and reading only about 450 bytes of it per word generated.
SK Hynix's Nasdaq listing raises $26.5 billion, the largest first-time US listing by a foreign company ever News
SK Hynix raised $26.5 billion in a Nasdaq listing that topped Alibaba's record for the largest US IPO ever by a foreign company, priced on the strength of its dominance in the high-bandwidth memory that AI accelerators depend on.
LaMem-VLA gives robots a working memory so they stop forgetting what they just did News
A new robotics model, LaMem-VLA, fixes the 'short-horizon bias' that makes robots react only to what's in front of them by weaving compressed memories of past actions directly into the model's internal representation -- so it can handle multi-step tasks that require remembering an earlier state.
LaMem-VLA gives robots a memory so they stop forgetting the task News
A new framework called LaMem-VLA tackles the 'goldfish memory' problem in robot policies by compressing past experience into latent memory tokens and weaving them into the robot's current reasoning, targeting long-horizon manipulation tasks that single-frame models fail.
What should an AI agent remember about you, and what leaks when it does? News
Researchers are asking whether AI agents are ready for real long-term memory, just as another study shows how much an agent's memory can quietly give away about the people it served.
What does your AI actually remember about you? News
Two new studies stop trusting that agent 'memory' works and start measuring it directly, with results that carry a privacy sting.
Agent memory: how an AI remembers you after the conversation ends Lesson
Why most AI assistants have amnesia, the difference between short-term context and real long-term memory, and why remembering you is both what makes agents useful and what makes them a privacy risk.
What is a context window? Lesson
A model's context window is how much text it can hold in mind at once — its working memory. Bigger is useful, but a long window isn't the same as a good memory. Here's how it works and where it breaks.
AI 'world models' have short-term memory — they forget what's off-screen News
A sweeping study of dozens of AI video-prediction systems finds they don't truly remember the world; when something leaves the frame, they quietly reinvent it the next time you look.
cognee Tool
An open-source persistent-memory layer for AI agents with remember, recall, forget, and improve operations over a graph-plus-vector store, able to run graph relations, embeddings, session cache, and metadata in a single Postgres instead of four services.
TencentDB Agent Memory Tool
MIT-licensed memory layer that compresses conversation history into a semantic hierarchy of atoms, scenarios, and personas, with a gateway exposing capture, search, and recall endpoints. Its own benchmarks report token savings in the 31 to 61 percent range. Note that bearer auth and CORS allow-listing both default to off.
ChatGPT Computer History Tool
An opt-in feature in the ChatGPT desktop app on macOS that turns activity across allowed apps and websites into a searchable timeline ChatGPT and Codex can reference, and can surface repeated workflows as suggested skills or automations. Off by default, no screenshots or audio, temporary event files deleted after 48 hours. OpenAI's own docs warn it increases prompt-injection risk.