HP this week unveiled HP IQ, a new workplace intelligence layer designed to give business laptops enhanced productivity features. The software runs on a local artificial intelligence model, relying on OpenAI's gpt-oss-20b to power document analysis, chat functions, and automated meeting summarisation without relying on cloud infrastructure.
The appeal is evident. Processing data locally rather than sending it to external servers appeals to organisations concerned about data security and recurring API costs. HP IQ is designed to help coordinate experiences across HP AI PCs and workplace devices, bringing together local, on-device AI, proximity-based connectivity, and enterprise manageability to help reduce friction and keep work moving across devices and spaces.
HP IQ-enabled devices are expected to begin shipping in autumn 2026, with broader availability rolling out across the HP portfolio in the second half of 2026. Early access will require at least 24 GB of RAM on compatible EliteBook and ProBook models.
Beyond documentation and analysis, however, the product's meeting recording capability presents genuine complications. HP IQ can capture in-person meetings using laptop microphones, then generate summaries and action items. The problem is transparency. The local 20B-parameter model is designed to support responsiveness and privacy. Yet nowhere in HP's product description does the company mandate that participants be notified during recording, nor does it provide a clear visual indicator when capture is active.
This matters considerably under Australian law. In Australia, workplace laws and the Surveillance Devices Act require employers to navigate carefully around issues of privacy and consent when recording conversations or meetings. The Surveillance Devices Act 2007 in NSW prohibits anyone from making an audio recording of a private conversation when the person doing the recording is not a party to the conversation, or is a party but does not have the consent of all other parties. While some states like Queensland allow single-party recording, the Fair Work Commission has repeatedly found that secretly recording conversations at work strikes at the heart of the employment relationship and undermines the necessity of trust and confidence in the employment relationship.
When raised about the privacy implications, HP acknowledged best practice recommends obtaining permission before recording coworkers. Yet the company's own demonstration raised no onscreen notification or consent request; viewers could not tell from the laptop display that recording was occurring. HP confirmed the tool does not store audio recordings or generate full transcripts, features that might actually be useful for transparency purposes.
The other major component, HP NearSense, uses Wi-Fi, Bluetooth, and microphone signals to detect nearby devices and enable drag-and-drop file sharing between colleagues in the same room. This represents a practical convenience; it mirrors Apple's long-established AirDrop feature for wireless ecosystems. The technology also allows single-click joining to meeting rooms equipped with HP Poly conferencing hardware when users are physically present.
These productivity gains address a real problem. IT analysts note that smaller businesses and non-technical users struggle to implement local AI safely and effectively. HP IQ runs on a local language model with 20 billion parameters, with the vast majority of processing happening on the device itself; only when company policy and user rights allow it are specific tasks routed to the cloud. This approach avoids the substantial per-query costs of cloud-based AI services and keeps sensitive company information off external servers.
The broader shift toward open-source, local models does represent meaningful progress. OpenAI's GPT-OSS series is their first open-weight release since GPT-2, with two models (gpt-oss-20b and gpt-oss-120b) licensed under Apache 2.0, so developers can run them locally, fine-tune them, and even use them commercially without legal restrictions. This gives organisations genuine flexibility compared to proprietary cloud AI locked behind subscription fees and data-sharing agreements.
Yet HP's implementation suggests the technology industry has not fully solved the governance problem. Workplace productivity tools that incorporate recording capabilities require explicit consent mechanisms built into the user interface, not optional best practices recommended after launch. Australian employment law demands this transparency, and employee trust depends on it. HP should ensure any meeting recording feature includes an unmistakable on-screen indicator, requires affirmative consent before capture begins, and provides clear controls for disabling the feature.
The device will likely appeal to small to medium enterprises seeking on-premises AI without the complexity and cost of cloud infrastructure. The real question is whether HP will update its consent and notification protocols to match Australian workplace expectations before broader rollout. Without those safeguards, even a technically sound and productive tool will face resistance from compliance-conscious employers and cautious employees.