SecurityBrief New Zealand - Technology news for CISOs & cybersecurity decision-makers
New Zealand
Plaud urges Australian employers to tighten AI privacy

Plaud urges Australian employers to tighten AI privacy

Tue, 17th Nov 2026 (Today)
Joseph Gabriel Lagonsin
JOSEPH GABRIEL LAGONSIN News Editor

Plaud has urged Australian employers to make privacy and consent central to workplace AI use as AI assistants become more common in meetings and day-to-day office work.

Megumi Yoshinaga, Head of APAC & MEA Growth at Plaud, said organisations should focus not only on what AI tools can do, but also on what information they capture, where it is stored, who can access it and whether it can be deleted.

Her comments focus on the growing use of AI note-taking and transcription tools in workplaces, where conversations can include customer details, personal data, intellectual property, commercial discussions and confidential plans. These tools can reproduce sensitive information quickly and at scale, she said, raising governance questions as well as security concerns.

A recurring concern is the use of personal AI accounts for work conversations. Yoshinaga said this can leave transcripts and company knowledge spread across personal accounts or cloud folders with uneven access controls.

Trust and control

Plaud argues that workplace AI discussions in Australia are entering a new phase, with trust becoming as important as functionality. Yoshinaga said early adoption often focused on speed and convenience, such as generating content, summarising meetings and automating tasks. As those uses become routine, workers and employers are paying closer attention to how information is handled after it is captured, she said.

"Trust is becoming even more crucial as AI adoption accelerates," said Yoshinaga, Head of APAC & MEA Growth at Plaud.

As different AI tools begin to offer similar outputs, organisations are likely to look beyond product features, she said. In that environment, privacy, security and user control will play a larger role in procurement and deployment decisions.

Consent is another issue Plaud raised. Yoshinaga said employees and meeting participants should not be left guessing whether a conversation is being recorded or transcribed by AI. Instead, employers should build consent into normal workflows, she said.

"People should not have to wonder whether their conversation is being captured or processed by AI," Yoshinaga said.

Meaningful consent should be simple and tied to the actual use of the information, she said. In practice, that means participants should understand what is being captured, why it is being captured, who may access it and what happens to it afterwards.

Ownership questions

Plaud also highlighted data ownership and deletion rights. Yoshinaga said organisations should be able to identify where information created by AI assistants resides and who has the authority to remove it.

"Organisations need to be able to answer a basic question: where does the information live, and who can delete it? If they cannot answer that, they don't own it. They're renting it," Yoshinaga said.

She compared weak control over AI-generated records to using a rental car, where someone else keeps the spare keys and decides how the asset is managed. Her argument is that transcripts created in meetings become company records as soon as they are produced, and may carry obligations around access, retention, security and deletion.

The issue becomes more complex because the person who starts a recording is not necessarily the only person whose information is captured. Plaud said companies need clear internal rules covering control, sharing and disposal of AI-generated notes and transcripts.

Provider role

Yoshinaga said technology providers also have a responsibility to make privacy choices understandable. Settings should not be hidden behind technical complexity, she said, and users should be able to make informed choices without specialist knowledge.

"Technology providers have an important role in making privacy understandable and giving users meaningful control," Yoshinaga said.

Plaud said its approach includes a default policy under which user content is not used to train its own or third-party AI models unless a user has given explicit, informed consent in advance. It also said user data is encrypted and stored on regional cloud infrastructure based on the user's location.

For employers considering AI note-taking tools, Yoshinaga said they should examine what information is captured, where it is processed and stored, whether it is used to train models, how long it is retained and how much control users or organisations have over deletion. She also pointed to centrally managed workspaces as one way for businesses to gain clearer oversight of users, devices, subscriptions and collaboration.

"When every product can summarise a meeting, a key question is whether you would trust it in the room," Yoshinaga said.