Acceptable Use Policy

1. Lawful content only

Use Glassa only for lawful work. Don't create, store, send or share anything in Glassa that is unlawful, or that:

• sexually exploits or abuses children;
• threatens or harasses anyone, or incites violence or hatred against them;
• promotes terrorism or violent extremism;
• discriminates against people unlawfully;
• infringes someone else's copyright, trade marks, privacy or other rights;
• defames someone, or deceives people to their harm, such as fraud.

2. Mail

These rules apply to all mail sent from Glassa, by a member or by a teammate, whatever address it is sent from:

• Don't send spam, or bulk or marketing mail to people who did not ask for it, and don't mail lists you bought or scraped.
• Follow the law on electronic marketing wherever your recipients are, such as the UK's Privacy and Electronic Communications Regulations and the US CAN-SPAM Act: say who you are, and give every recipient of marketing mail a simple way to opt out.
• Don't disguise who a message is from: no forged headers, and no mail that pretends to come from another person or organization.
• Mail a teammate sends must say that it comes from an AI teammate of your organization, and must not pass as written by a person.
• Don't send phishing, or any mail that tries to trick someone into giving away a password, payment details or other personal data.
• Stop mailing anyone who asks you to.

3. No attacks on Glassa or anyone else

• Don't try to get into another organization's workspace, another member's account, or any part of Glassa you have not been given access to.
• Don't probe, scan, load-test or test the security of Glassa without our written permission, and don't get around its access controls, rate limits or plan limits.
• Don't upload or send malware, and don't use Glassa to attack, scan or flood any other system.
• Don't reverse-engineer Glassa, except where the law allows it.

To test Glassa's security, write to hello@easyloops.ltd first, and test only what we agree in writing. If you find a security weakness in Glassa, tell us at the same address, and don't use it or share it with anyone else.

4. AI teammates

Glassa shows a teammate as AI in your organization's people directory and on its profile, and names it "Agent" followed by a name, which you can change. Everything in this policy applies to what a teammate does, as it does to a person, and in addition:

• Don't use a teammate to pass as a person, or to impersonate any real person, in chat, calls, meetings, mail or anywhere else.
• Don't give a teammate a name or picture that hides that it is AI, and don't tell anyone it is a person.
• Make sure that everyone a teammate deals with knows it is AI, especially people who may not see how Glassa marks it, such as people in a call or meeting with it and anyone outside your organization who receives its messages, mail or work.
• Don't use a teammate to do anything this policy does not allow you to do yourself, or to get around the access your organization has given it.

5. Prohibited AI practices

Don't use Glassa, its teammates or its AI features, such as meeting transcripts and notes, for any practice that Article 5 of the EU AI Act prohibits, wherever you are. These include:

• inferring the emotions of people at work or in education, for example from their voice or face in a meeting, except for medical or safety reasons;
• sorting people by their biometric data to infer their race, political opinions, trade union membership, religious or philosophical beliefs, sex life or sexual orientation;
• manipulating or deceiving people with techniques that distort their behaviour and are likely to cause them significant harm;
• exploiting people's vulnerabilities due to their age, disability, or social or economic situation;
• social scoring: rating people on their social behaviour or personal traits in a way that leads to unjustified or unfair treatment of them;
• predicting that a person will commit a crime solely from profiling them;
• building facial recognition databases by untargeted scraping of facial images from the internet or CCTV.

6. High-risk uses

Glassa's teammates and AI features are not designed or assessed for the uses the EU AI Act treats as high-risk. Wherever you are, don't use them to make decisions about people, or to score, rank, filter or profile people, in any of these areas:

• recruitment and work: choosing who to hire or promote, ending someone's work, allocating work by people's behaviour or personal traits, or monitoring and assessing how people perform or behave at work;
• education: deciding who is admitted, assessing learners, or watching them during tests;
• essential services: deciding whether people get public benefits or services, credit, or life or health insurance, or on what terms, or prioritizing emergency calls;
• identifying people from their biometric data, such as their face or voice, at a distance, or sorting people by sensitive traits inferred from it;
• the safety of critical infrastructure, such as roads, water, gas, heating or electricity;
• law enforcement, migration, asylum and border control, courts and justice, or elections.

Notes of a meeting, such as an interview, may record what was said. Any decision about the people in it must be made by a person, on their own assessment.

7. Recording and transcribing conversations

Glassa can record what is said in a meeting or call to write a transcript and notes, and a teammate in a call hears and transcribes what is said. Before you turn on notes, or bring a teammate into a call:

• Tell everyone in the conversation, including guests and anyone you can hear on a shared device or phone, that it is being recorded and transcribed, and that a teammate is there if one is.
• Get any consent the law requires. In some places one person in a conversation may record it; in others, including several US states, everyone in it must agree. Follow the strictest law that applies to anyone taking part.
• Don't record or transcribe anyone in secret, or a conversation you are not part of, and stop when someone who has the right to object does so.

8. Special category data

Glassa is not designed for special category personal data: data about a person's health, racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, sex life or sexual orientation, genetic data, or biometric data used to identify them. Don't put any in Glassa unless your organization has its own lawful basis, and a condition under data protection law, for processing it, and has recorded both. The same applies to data about criminal convictions and offences.

9. Sanctions and export controls

Don't use Glassa, and don't let anyone use your workspace, if you, your organization or anyone who owns or controls it is the target of sanctions imposed by the UK, the United Nations, the EU or the US, or is in a country or region under comprehensive sanctions. Don't use Glassa to send, share or give access to software, technology or data in breach of export control laws, including those of the UK and the US.

10. When this policy is broken

We act on what we find or are told, in proportion to the harm. We may:

• remove or block content, or stop mail from being sent;
• pause a teammate, or limit a member's or the workspace's mail;
• suspend a member's account or the workspace;
• end your organization's access for a serious or repeated breach, as our Terms of Service allow.

We normally tell your organization's owners first and give them time to put it right. When the harm is serious or urgent, we act first and tell them after, and we tell them nothing when the law forbids it. We report content or activity to the police or other authorities where the law requires us to, and we always report child sexual abuse material. If you think we got it wrong, reply to our notice or write to hello@easyloops.ltd, and we will look at it again.

11. Reporting misuse

To report misuse of Glassa, such as spam sent from it, write to hello@easyloops.ltd with what you saw. For mail, include the message with its full headers.

12. Changes

If we change this policy in a way that matters, we will publish it as a new version, with the date it takes effect above, and tell organizations in Glassa or by email at least 30 days before the change applies.